Microphone or line source - VISER-u
The Sound Studio
I dedicate this Seventh Edition to my son Guy
rather belatedly, since he trod his own distinctive path
towards the virtual studio and can now tell me a few
things about what goes on there. I guess he doesn't
really need this book ± but it's the thought that counts.
The Sound Studio
Audio techniques for radio, television, film and recording
Seventh edition
Alec Nisbett
Focal Press
An imprint of Elsevier
Linacre House, Jordan Hill, Oxford OX2 8DP
200 Wheeler Road, Burlington MA 01803
First published 1962
Second edition 1970
Third edition 1972
Fourth edition 1979
Fifth edition 1993
Sixth edition 1995
Reprinted 1995, 1999, 2000
Seventh edition 2003
Copyright # 1962, 1970, 1972, 1979, 1993, 1995, 2003, Alec Nisbett. All rights reserved
The right of Alec Nisbett to be identified as the author of this work
has been asserted in accordance with the Copyright, Designs and
Patents Act 1988
No part of this publication may be reproduced in any material form (including
photocopying or storing in any medium by electronic means and whether
or not transiently or incidentally to some other use of this publication) without
the written permission of the copyright holder except in accordance with the
provisions of the Copyright, Designs and Patents Act 1988 or under the terms of
a licence issued by the Copyright Licensing Agency Ltd, 90 Tottenham Court Road,
London, England W1T 4LP. Applications for the copyright holder's written
permission to reproduce any part of this publication should be addressed
to the publisher
Permissions may be sought directly from Elsevier's Science and Technology Rights
Department in Oxford, UK: phone: (+44) (0) 1865 843830; fax: (+44) (0) 1865 853333;
e-mail: [email protected] You may also complete your request on-line via the
Elsevier homepage (www.elsevier.com), by selecting `Customer Support'
and then `Obtaining Permissions'
British Library Cataloguing in Publication Data
Nisbett, Alec
The Sound studio: audio techniques for radio, television, film and recording. ± 7th ed.
1. Sound ± Recording and reproducing
I. Title
621.30 893
Library of Congress Cataloguing in Publication Data
A catalogue record for this book is available from the Library of Congress
ISBN 0 240 51911 6
For information on all Focal Press publications visit our website at:
www.focalpress.com
Typeset by Integra Software Services Pvt. Ltd, Pondicherry, India
www.integra-india.com
Printed and bound in Great Britain
Contents
Introduction
About the author
1 Audio techniques and equipment
Studio operations
The sound control room
Studio layout for recording
Studio communications
2 The sound medium
The nature of sound
Wavelength
Waves and phase
Waves and obstacles
Energy, intensity and resonance
Overtones, harmonics and formants
Air resonance
The voice
The human ear
Sound volume and the ear
Loudness, frequency and hearing
3 Stereo
Two loudspeakers
Reduced loudspeaker separation
Alternatives to two loudspeakers
Hearing and stereo
Microphones for stereo
Stereo compatibility
ix
xi
1
2
3
5
6
9
9
11
12
13
14
15
18
20
21
21
22
25
25
27
27
29
30
33
4 Studios and acoustics
The range and function of studios
Noise and vibration
Reverberation
Coloration
Studios for speech
General-purpose sound studios
Music studios
Dead pop music studios
Television and film studios
Acoustic treatment
Using sound absorbers
The use of screens
Altering the acoustics of concert halls
Acoustic modelling of studios
Acoustic holography
5 Microphones
Microphone properties
Directional response
The frequency response of practical
microphones
Proximity effect
Cardioid and hypercardioid microphones
Switchable microphones
Highly directional microphones
Noise-cancelling microphones
Microphones for use in vision
Boundary microphones
Personal microphones
Radio microphones
Contact and underwater microphones
34
34
36
37
38
39
40
41
42
43
44
46
47
48
50
53
54
55
57
59
61
63
65
66
70
72
74
75
77
79
vi
Contents
Directional characteristics of A and B
stereo pairs
Windshields
Mountings
Cables and connectors
Microphone checks
Lining up a stereo pair
6 Microphone balance
Stereo microphone balance
The balance test
Visual conventions for microphones
Multimicrophone balance
Artistic compatibility: mono, stereo
and surround balances
7 Speech balance
Microphones for speech
Interviewing with a hand-held microphone
Three or more voices
Studio noise
Handling a script
Speech in stereo
Boom operation
Slung and hidden microphones
Microphones in vision
Using gun microphones
Using personal microphones
Radio drama
Open-air acoustics
Stereo drama
Stereo drama problems
Audience reaction
8 Music balance
`Natural' balance
Music studio problems
One microphone or many?
Individual instruments and groups
Violin, viola
Cello, bass
More strings: acoustic balances
Electric guitar family
Grand piano
Piano and soloist, two pianos
Upright piano
Harp, harpsichord, celesta
Piano accordion
Harmonica
80
82
83
85
88
88
90
91
92
95
96
97
99
9
99
102
103
104
104
105
106
108
108
110
110
112
113
115
116
117
Woodwind
Saxophones
Brass
Percussion, drums
Singers: solo and chorus
The orchestra
Orchestra: spaced microphones
Pipe organ with orchestra
Orchestra with soloists or chorus
Opera
Classical music in vision
Televised opera
Popular music
Larger groups
Multimicrophone layout
Popular music in vision
134
135
136
138
140
141
144
144
145
145
147
149
150
152
153
154
Monitoring and control
156
Quality and the ear
Loudspeakers
Monitoring layout
Control consoles and racks
Digital conversion
Automated and digital consoles
Channels
Group and master controls
Faders
Stereo image controls
Tracklaying, multitrack recording and
mix-down
Control console checks
Monitoring sound quality: noise
Hum
Distortion
Crosstalk
Subjective quality checks
156
157
159
161
163
166
167
169
170
171
173
174
175
177
177
179
179
Volume and dynamics
180
Meters
Line-up tone
Multitrack meters
Controlling stereo
Programme volume: relative levels
Maximum volumes
Dynamic range and compression
Compression of music
Compression: speech and effects
Compressors and limiters
Compressor design and use
`Gating' background noise
181
183
185
186
187
188
189
190
191
192
193
194
119
120
121
123
123
124
126
127
128
129
131
132
133
134
134
10
Contents
11
12
13
14
15
Filters and equalization
196
Filters
Graphic equalizers
Telephone effects
197
199
200
Reverberation and delay effects
203
AR feeds in the console
Real reverberation
Analogue AR
Digital AR
Using AR on speech
Delay-related effects
Changing pitch
Sampling
204
205
207
208
210
210
212
214
Recorded and remote sources
216
Live-and-insert programmes
Preparing recorded inserts
Cueing play-in
Remote sources
Clean feed and multi-way working
Telephone circuits in radio and
television programmes
217
218
220
223
225
Fades and mixes
229
The fade in radio drama
Mixing in effects
Fades in television and film
Theme and background music
Mixing from speech to music
Joining music to music
Adding speech to music
229
232
233
235
236
237
238
16
17
18
226
Sound effects
239
The conventional use of effects
Surrealistic effects
Spot effects
Doors
Footsteps
Telephones, bells and buzzers
Personal action effects
Gunplay
Creaks, squeaks, swishes, crackles,
crashes and splashes
Horses' hooves
The recorded sound picture
The use of recorded effects
Recorded effects in stereo
Changing the pitch of an effect
The recorded effects library
Acoustic foldback
Effects for television and film
240
242
243
244
245
247
248
249
251
252
253
254
256
256
257
258
258
19
20
vii
Footsteps in vision
Wind, rain and fire in vision
Sound equipment in vision
Television sports and events
Sound effects on film
259
260
261
261
262
The virtual studio
264
MIDI plus audio
Plug-ins
Control surfaces
Digital problems
266
267
270
270
Shaping sound
272
Musique concreÁte
Electronic sources
New laws for music
Animated sound
274
276
277
279
Audio editing
281
Retakes and wildtracks
Tape editing
Rough editing
Fine editing
Finding the exact point
Editing sound effects
Editing music
Hard disk editing options
Digital projects and editing
Play-out systems
283
284
286
287
288
290
290
291
293
296
Film and video sound
298
Time code
Picture and sound standards
Choice of location
Film and video sound recording
Film and video recording routine
Film and video documentation
Videotape editing
Digital video editing
Rough and fine cutting
Shot list, commentary and music
Tracklaying
Conforming sound and picture
Post-synchronization
Mixing sound for picture
Film sound editing
299
300
302
303
304
306
308
308
311
312
314
315
316
317
320
Planning and routine
325
Planning: radio or recordings
Television planning
The television director
325
326
327
viii
Contents
The television designer
Planning microphone coverage
Preparations for rehearsal
Line-up
Production timing
Cueing methods
Cueing applause
Documentation
328
330
331
332
333
334
335
337
21 Communication in sound
339
Sound and the imagination
The audience
340
341
If there's no picture ± where is everybody?
The time factor
Understanding and remembering
The communication of ideas
Pace
342
344
346
348
349
Glossary
351
Bibliography
377
Index
379
Introduction
The audio revolution is over ± digital is king! One of my principal
aims for this Seventh Edition has been to make sure that current
practice in digital audio is fully represented. This will include such
topics as: the virtual studio, along with aids like Pro Tools and
reattached MIDI keyboards; 5.1 surround sound; hard drive mixers
and multichannel recorders, plus DVD and CD-RW and . . . you
name it (although on second thoughts you may not have to: computers seem to have a whole naming industry to support them).
And yet . . . a curious thing happened. I found that in the first eight
chapters (and in several of the others) there was surprisingly little to
say about this new digital world we all live in. So ± why was that?
The short answer is simply that sound is not digital. Your ears and
voice are not digital. The way that your brain processes sound is
rather odd ± but not digital. Nor are the loudspeakers you listen to, or
(even more to the point) the microphones you set up to capture
sound. If you encounter something called a digital microphone, what
that really means is that the analogue electrical signal generated
within it is turned into digital information before being sent down
a cable to the digital processor or recorder downstream . . . but it will
still have to be converted back to analogue sound before you can
hear it. What we really live in today is a hybrid analogue-and-digital
world. Some of the analogue parts are unavoidable; the digital bits
are optional.
Where digital processes actually can be used they are often better.
Digital information can be passed over greater distances and through
more pieces of equipment that do things to it (process it), with far
smaller losses of information along the way. Digital recording is
more faithful, and digital recording media can hold ever more
information. From the outside that might seem like revolution
enough, but even more profound are the increases in processing power
and processing speed that mean more can be accomplished (and
x
Introduction
completed in real time) than was possible before. This change is not
just quantitative, it is qualitative: in this new digital world you can
do things that were not possible, or not practicable, before.
This does not mean that all processes or processing have to be digital
all the time. Many people who use digital processes most of the time
like to return occasionally to older analogue devices or manual
techniques in order to experience their `hands-on' qualities more
directly. Or, for a particular application, they replace smoothly
operating new technology by clunky old gear, to bring back some
perceived special sound quality that came with it. I am not a great
fan of this aspect of `retro', as it may be called; after all, in this new
digital world you can always simulate any older kind of `clunky'
alongside an endless range of the new.
There is another reason why, in this book, I still describe some
apparently archaic techniques. It is because many operations are
designed to do the same thing whether they are achieved by analogue or digital means. Often, the terminology is unchanged and can
be understood better in its original context, even though a computer
may now do it for you.
In this book I will also make some short cuts ± by, for example,
putting in very little about particular techniques that have been
painstakingly developed and exhaustively analysed, even to become
mainstream for some of their users, when there are competitive
techniques that in my view do the same job as well or better, and
seem to have wider application. An example is `dummy head' and
related stereo microphone techniques. You may meet (or even work
alongside) people who swear by these: if so, they will direct you
to books that describe them and explain their justification in
more detail.
All this remains consistent with the original purpose of this book: to
explain the subject simply, in terms that are acceptable both to the
creative people in production ± including writers and performers ±
and to those who deal operationally with the balance and recording
of sound or with its engineering aspects. By exploring the ground
common to all these groups, the work of each may be interpreted to
the other. It is also designed for those who may want to explore
audio techniques that have been developed in fields that are remote
from their own, whether this be radio, music recordings, television
or film sound.
In these aims I am aided by the same digital advances that now
pervade audio. This Seventh Edition is in part a computer construct,
derived initially by optical character recognition from the Sixth
Edition, then revised topic by topic and scrutinized and updated
paragraph by paragraph in far greater detail than ever before. In
consequence, this new version should be not only more finely
attuned to current practice, but also (I hope) even more readable.
Several features of this book remain almost unchanged in this edition. One is its structure and the underlying logic, with each chapter
in turn introducing ideas that are used later throughout the book.
Another is that despite all the technical advances in equipment that
Introduction
xi
are touched on, it remains a book about audio techniques, not audio
engineering. No previous scientific, mathematical or engineering
knowledge is assumed.
Since its first edition, my emphasis has always been on general
principles rather than rules of thumb, though the shorter routes to
high-quality results are not neglected. I still insist on the paramount
importance of subjective judgement, of learning to use our ears, of
exercising well-developed aural perception, allied to a strong critical
faculty. Throughout, I describe studio and location techniques in
terms of what each operation does to the sound and why, and what to
listen for.
Of all the chapters in this book, that which has changed least is the
last, on communication in sound. Happily, in the years since it was
first written, neither scientific study of this human capacity nor the
evolution of the media themselves has invalidated its observations.
But some reviewers have suggested that I have put this last chapter
in the wrong place: they strongly suggest that you read it first!
About the author
If you want to know what prejudices to expect, or why the advice or
terminology here may be a little different from that in some other
book you have seen, it may help you to know where the writer is
coming from. So . . . this is where I came from.
I began my working life as a studio manager in BBC Radio ± for
which they did not want `engineers'. In fact, the entry qualifications
seemed to be `anything unusual, but different from the last successful applicant', of which about 50% were women and a few were gay.
Oddly, being a male mathematical physicist did not seem to disqualify me. My earliest faltering footsteps (the `spot effects' later
called `foley' by others) included horses' hoof effects generated with
the aid of coconut shells. I went on to learn and use more of the
basics ± such as `balance' ± that you will read about here. I also
helped to outline what would be needed for a `Radiophonic Workshop' ± now gone, but an early precursor to the virtual studios that
generate much of today's music.
Later I moved to television, making both studio productions and
`outside broadcasts' (as we called them). I became a filmmaker and
specialized in science documentaries such as Horizon (the British
precursor to the American Nova series on PBS) of which I made 42,
probably more than anyone else alive. Throughout all this, I maintained a particular concern for the audio input: voice, effects, and
both conventional and synthesized music. I led co-productions with
lots of other broadcasters in many other countries, collected a bunch
of the usual awards and have long been a life member of the British
Academy of Film and Television Arts.
Wee have also Sound-Houses, wher wee practise and demonstrate
all Sounds, and their Generation. Wee have Harmonies which you
have not, of Quarter-Sounds, and lesser Slides of Sounds. Diverse
Instruments of Musick likewise to you unknowne, some Sweeter then
any you have; Together with Bells and Rings that are dainty and
sweet. Wee represent Small Sounds as Great and Deepe; Likewise
Great Sounds, Extenuate and Sharpe; Wee make diverse Tremblings
and Warblings of Sounds, which in their Originall are Entire. Wee
represent and imitate all Articulate Sounds and Letters, and the
Voices and Notes of Beasts and Birds. Wee have certaine Helps,
which sett to the Eare doe further the Hearing greatly. Wee have also
diverse Strange and Artificiall Eccho's, Reflecting the Voice many
times, and as it were Tossing it; And some that give back the Voice
Lowder then it came, some Shriller, and some Deeper; Yea some
rendring the Voice, Differing in the Letters or Articulate Sound,
from that they receyve. Wee have also meanes to convey Sounds in
Trunks and Pipes, in strange Lines, and Distances.
From The New Atlantis by Francis Bacon
1624
Chapter 1
Audio techniques and
equipment
If you work in a radio or recording studio, or with television or film
sound; if you are in intimate daily contact with the problems of the
medium; if you spend hours each day consciously and critically
listening, you are bound to build up a vast body of aural experience.
Your objective will be to make judgements upon the subtleties of
sound that make the difference between an inadequate and a competent production, or between competent and actively exciting ± and
at the same time to develop your ability to manipulate audio equipment to mould that sound into the required shape.
The tools for picking up, treating and recording sound continue to
improve, and it also becomes easier to acquire and apply the basic
skills ± but because of this there is ever more that can be done.
Microphones are more specialized, creating new opportunities for
their use. Studio control consoles can be intimidatingly complex, but
this only reflects the vast range of options that they offer. Using
lightweight portable recorders or tiny radio transmitters, sound can
be obtained from anywhere we can reach and many places we
cannot. By digital techniques, complex editing can be tackled. In
fact, home computers can now simulate many of the functions
of studios.
But however easy it becomes to buy and handle the tools, we still
need a well-trained ear to judge the result. Each element in a
production has an organic interdependence with the whole; every
detail of technique contributes in one way or another to the final
result; and in turn the desired end-product largely dictates the
methods to be employed. This synthesis between means and end may
be so complete that the untrained ear can rarely disentangle them ± and
even critics may disagree on how a particular effect was achieved.
This is because faults of technique are far less obvious than those of
2
Audio techniques and equipment
content. A well-placed microphone in suitable acoustics, with the
appropriate fades, pauses, mix and editing, can make all the difference to a radio production ± but at the end an appreciative listener's
search for a favourable comment will probably land on some remark
about the subject. Similarly, where technique is faulty it is often,
again, the content that is criticized. Anyone working with sound
must develop a faculty for listening analytically to the operational
aspects of a production, to judge technique independently of content,
while making it serve content.
In radio, television or film, the investment in equipment and the
professionals' time makes slow work unacceptable. Technical problems must be anticipated or solved quickly and unobtrusively, so
attention can be focused instead on artistic fine-tuning, sharpening
performance and enhancing material. Although each individual concentrates on a single aspect, this is geared to the production as a
whole. The first objective of the entire team is to raise the programme to its best possible standard and to catch it at its peak.
This first chapter briefly introduces the range of techniques and
basic skills required, and sets them in the context of the working
environment: the studio chain or its location equivalent. Subsequent
chapters describe the nature of the sound medium and then elements
of the chain that begins with the original sound source and includes
the vast array of operations that may be applied to the signal before it
is converted back to sound and offered to the listener ± who should
be happily unaware of these intervening stages.
Studio operations
The operational side of sound studio work, or for sound in the visual
media, includes:
.
.
.
.
.
.
.
.
Choice of acoustics: selecting, creating or modifying the audio
environment.
Microphone balance: selecting suitable types of microphone and
placing them to pick up a combination of direct and reflected
sound from the various sources.
Equalization: adjusting the frequency content.
Feeding in material from other places.
Mixing: combining the output from microphones, replay equipment, devices that add artificial reverberation and other effects,
remote sources, etc., at artistically satisfying levels.
Control: ensuring that the programme level (i.e. in relation to the
noise and distortion levels of the equipment used) is not too high
or too low, and uses the medium ± recording or broadcast ±
efficiently.
Creating sound effects (variously called foley, hand or spot effects)
in the studio.
Recording: using tape, disc and/or hard disk (including editing
systems and back-up media) to lay down one, two or more tracks.
Audio techniques and equipment
.
.
.
.
.
3
Editing: tidying up small errors in recordings, reordering or
compiling material, introducing internal fades and effects.
Playing in records and recordings: this includes recorded effects,
music discs or tapes, interviews and reports, pre-recorded and
edited sequences, etc.
Feeding output (including play-out): ensuring that the resultant
mixed sound is made available to its audience or to be inscribed
upon a recording medium with no significant loss of quality.
Documentation: keeping a history of the origins, modification
and use of all material.
Storage: ensuring the safety and accessibility of material in use
and archiving or disposing of it after use.
In small radio stations in some countries, a further responsibility
may be added: supervising the operation of a transmitter at or near
the studio. This requires certain minimal technical qualifications.
The person responsible for balancing, mixing and control may be a
sound or programme engineer, or balance engineer (in commercial
recording studios); in BBC radio is called a studio manager; in
television may be a sound supervisor; on a film stage is a sound
recordist (US: production sound recordist); or when re-recording
film or video, a re-recording mixer (US) or dubbing mixer (UK).
Although in many parts of the world the person doing some of these
jobs really is an engineer by training, in broadcasting ± and for some
roles in other media ± it is often equally satisfactory to employ men
and women whose training has been in other fields, because
although the operator is handling technical equipment and digital
operations, the primary task is artistic (or the application of craftsmanship). Real engineering problems are usually dealt with by a
maintenance engineer.
[1.1] Digital studio chain
1, Microphones. The electrical signal
from a microphone is analogue, so it
has to be fed through an analogue-todigital (A/D) converter, after which it
remains digital throughout the chain.
2, Digital console (mixer).
An alternative is a personal computer
(often, for media applications, a Mac)
with audio software that includes a
pictorial representation of faders, plus
`plug-ins'. 3, Modular digital
multitrack eight-track recorders
(MDMs): these can be linked
together to take up to 128 tracks,
although 32 are generally more than
enough. Here the alternative is hard
disk recording. 4, Digital audio tape
(DAT) recorder. 5, Digital audio
workstation (DAW ± more software
on a personal computer). 6, CD-R
(or DVD-R) usually built in to the
computer. 7, The end-product on CD.
Whatever title may be given to the job, the sound supervisor has
overall responsibility for all sound studio operations and advises the
director on technical problems. Where sound staff work away from
any studio, e.g. on outside broadcasts (remotes) or when filming on
location, a higher standard of engineering knowledge may be
required ± as also in small radio stations where a maintenance
engineer is not always available. There may be one or several
assistants: in radio these may help lay out the studio and set microphones, create sound effects, and make, edit and play in recordings.
Assistants in television and film may also act as boom operators,
positioning microphones just out of the picture: this increasingly rare
skill is still excellent training for the senior role.
The sound control room
In the layout of equipment required for satisfactory sound control,
let us look first at the simplest case: the studio used for sound only,
for radio or recording.
4
Audio techniques and equipment
The nerve centre of almost any broadcast or recording is the control
console ± also known as desk, board or mixer. Here all the different
sound sources are modified and mixed. In a live broadcast it is here
that the final sound is put together; and it is the responsibility of the
sound supervisor to ensure that no further adjustments are necessary
before the signal leaves the transmitter (except, perhaps, for any
occasional automatic operation of a limiter).
The console, in most studios the most obvious and impressive
capital item, often presents a vast array of faders, switches and
indicators. Essentially, it consists of a number of individual channels, each associated with a particular microphone or some other
source. These are all pre-set (or, apart from the use of the fader,
altered only rarely), so that their output may be combined into
a small number of group controls that can be operated by a single
pair of hands. The groups then pass through a main control to
become the studio output. That `vast array' plainly offers many
opportunities to do different things with sound. Apart from the
faders, the function of which is obvious, many are used for signal
routing. Some allow modification of frequency content (equalization) or dynamics (these include compressors and limiters). Others
are concerned with monitoring. More still are used for communications. Waiting to be inserted at chosen points are specialized
devices offering artificial reverberation and further ways of modifying the signal. All these will be described in detail later.
In many consoles, the audio signal passes first through the individual strips (`channels'), then through the groups, then to the
main control and out to `line' ± but in some this process is partly
or wholly mimicked. For example, a group fader may not actually
carry the signal; instead, a visually identical voltage control amplifier (VCA) fader sends messages which modify the audio in the
individual channels that have been designated as belonging to the
group, before it is combined with all other channels in the main
control. In some consoles, the pre-set equalization, dynamics and
some routing controls are located away from the channel strips,
and can all be entered separately into the memory of a computer:
this greatly reduces the (visual) complexity and total area of
the layout.
[1.2] Components of the studio
chain
1, The sound source. 2, Direct and
indirect sound paths. 3, Noise:
unwanted sound. 4, Microphone.
5, Preamplifier. 6, Equalizer.
7, Reverberation, now usually digital.
8, Tape head or disc player. 9, External
line source. 10, Mixer. 11, Meter.
12, Recorder. 13, Loudspeaker:
sound reaches the ear by direct and
reflected paths. 14, Studio output.
In the extreme case, the actual signal does not reach the console at
all: the controls send their messages, via a computer, to racks in
which the real signal is modified, controlled and routed. What then
remains of an original console layout depends on the manufacturer's
judgement of what the customer will regard as `user friendly': it
should be possible for the majority of users to switch between
analogue and fully digital systems (and back again) without having
to go through a major course of re-education.
Apart from the microphones and control console, the next most
important equipment in a studio is a high-quality loudspeaker
system ± for a radio or recording studio is not merely a room in
which sounds are made and picked up by microphone, it is also
the place where shades of sound are judged and a picture is
created by ear.
Audio techniques and equipment
5
The main thing that distinguishes a broadcasting studio from any
other place where microphone and recorder may be set up is that two
acoustically separate rooms must be used: one where the sound may
be created in suitable acoustics and picked up by microphone, and
the other in which the mixed sound is heard. This second room is
often called the control room ± although in organizations where
`control room' means a line-switching centre it may be called a
`control cubicle' instead. Tape, disc and hard-drive recorders and
reproducers and editing systems, and the staff to operate them, are
also in the control area, as are the producer or director (in some cases
these two roles are combined) and the production secretary (who
checks running and overall timing, etc.).
In the simplest television studios, sound and picture are controlled
side by side, but more often there is an acoustically separate sound
room with much the same physical layout as it would have in a radio
studio (except that it will have picture monitors, including one for
that which is currently selected and one for preview). The television
sound supervisor really does need to listen to programme sound in
a separate cubicle because the director and his assistant must give
a continuous stream of instructions on an open microphone, which
means that their programme sound loudspeaker has to be set at
a lower level than is necessary for the judgement of a good balance,
and at times their voices obscure it. Their directions are relayed to
the sound control room on an audio circuit, but at a reduced volume.
Studio layout for recording
Units of measurement
Tape widths and speeds are
commonly described by their
Imperial measurements: `quarterinch tape', `71/2 inches per
second', and so on. This is a rare
exception to the general rule in
science and engineering that
measures are presented in metric
(or `SI') units. In this book Imperial
measures are usually given as
well, for those who still prefer to
use them. Measurements such as
tape speeds are precise, so
conversions must be given fairly
accurately, but in other cases
throughout this book, measures
are intended only as ballpark
figures, and conversions are
rounded up or down as anything
more precise would give a false
impression.
It is no good finding a recording fault ten minutes after the performer has left the studio. If possible, a recording should be monitored
from the recording itself as it is being made. On simple recordings
the operational staff can do both jobs, operating the recorder and
monitoring the output from a separate replay head. Another safeguard is backup: for example, a hard disk recording might be
backed up both within the computer (recorded automatically in
two places at once) and also on tape ± but even this provides no
guarantee that it was actually recorded anywhere, unless at least
one is checked. Each track of a multitrack (tape or digital) recording
cannot realistically be checked continuously, but a series of recordings can usually be retained for a while. When recording a
live public event the options are more limited still: it is vital not
only to have backup but also to listen continuously for any possible
recording problem.
When a recording made on analogue tape is monitored while it is
being recorded, there is a delay of perhaps one-fifth of a second that
corresponds to the 3 inches of tape (8 cm) between recording and
reproducing heads at 15 inches per second (in/s) (38 cm/s) or a delay
of two-fifths of a second at 7.5 in/s (19 cm/s). This kind of delay
makes it unsatisfactory to combine monitoring a recording with
any other job when the production has any degree of complexity.
6
Audio techniques and equipment
[1.3] Radio studio layout
1, Studio. 2, Lobby. 3, Double or triple
glazed window for acoustic separation.
4, Control console and monitoring
loudspeakers. 5, Producer's position
(ideally, on same line as balancer).
6, Recording and replay equipment.
7, Equipment racks.
One-fifth of a second can completely destroy timing. The perfect
fade on a piece of music, for example, may have to be accurate to
less than one-twentieth of a second. These requirements suggest that,
ideally, for complex recordings there might be a third acoustically
separate room in which a recording engineer can check the quality of
the recorded sound ± although backup is generally cheaper than the
cost of extra people.
Many productions are prepared (or rehearsed) and then recorded
section by section in random order. Each segment of a few minutes
is separately discussed, parts or all of it run through, then recorded
(together with any necessary retakes), and the results are edited
together later. In the simplest arrangement, segments are recorded
in the control room, then checked and edited (using headphones)
while the next piece is being prepared. Using digital (non-destructive) editing systems, material can even be rough-edited as the recording continues ± to mark in and out points, and delete preliminary and
intervening production instructions, gaps and errors ± to create a rough
assembly.
An increasingly rare alternative is more like a stage play or `live'
broadcast. This is based on the idea that the dramatic contour of a
complex work is likely to have a better shape if it is fully
rehearsed and then recorded continuously as a whole, leaving any
faults to be corrected by retakes. Studio time may be divided into
two parts, with the recordings limited to a relatively short session.
Studio communications
[1.4] Loudspeakers in a television
musical production
C, Sound control room. G, Gallery.
S, Studio. 1, Audience (public address)
system. 2, Foldback of recorded
sound effects. 3, Special feed of
vocalist's sound to conductor.
4, Special feed of rhythm group to
vocalist. 5, Remote studio.
In both radio (and other recording studios) and television there are
sound circuits that are used for local communication only.
Headphones can be used by performers to hear a separate talkback
circuit from the control room. Their use may range from full-scale
relay of instructions (as in television productions), through feeding
questions to an interviewer, to monitoring a feed of some event in
Audio techniques and equipment
7
order to give news of it. There may be both talkback (to the studio)
and reverse talkback. The latter is an independent feed from the
studio, often associated with a microphone-cut key controlled by the
presenter in order to get off-air and ask for information or instructions (or another simply to get off-air and cough).
When a studio microphone is faded up it is usually arranged that the
studio loudspeaker be cut off automatically (except for talkback
purposes during rehearsal). Headphones or an earpiece may be used
in the studio to hear talkback when the loudspeaker has been cut,
which may be just before a cue.
In television the complexity of the communication system is greater,
because there are so many more people involved, and also because
they need to occupy a range of acoustically separate areas. The
director speaks to the floor manager over a local radio circuit; the
floor manager hears the director on a single earpiece; camera operators and others in the studio listen on headphones.
In the sound control room the sound supervisor can hear high-level
and high-quality programme sound. In addition, a second, much
smaller loudspeaker (so that it is distinctly different in quality)
relays both the director's instructions and reverse talkback from
sound assistants on the studio floor.
Lighting and camera control also generally have a (third) acoustically separate area. They have their own talkback system to appropriate points on the studio floor, and camera operators in their turn
have reverse talkback to them. In the control gallery a technical
manager has, in effect, a small telephone exchange for communication with people at all points concerned with technical aspects of the
production. This includes control lines to remote sources, and also to
video recording and play-out and telecine operators (who, in their
own areas, hear the director's circuit as well as programme sound or
their own output).
Television outside broadcast or remote trucks sometimes have sound
control, director and camera control all in the same tiny area of the
vehicle: this is best for sports and other unrehearsed events where
side-by-side working helps the operators to synchronize their
actions. But for rehearsed events, and productions built on location
sequence by sequence, good sound conditions are more important,
and these demand an acoustically separate area. This may be either
in the same mobile control vehicle or (for music, where it is better to
have more space for monitoring) in a separate truck or, alternatively,
`derigged' and set up in some convenient room.
[1.5] Elements of studio chain
1, Microphones: low-level sources,
fed through amplifier. 2, High-level
sources (replay machines, etc.).
3, Microphone and line-input channels
in control console. 4, Group channels
in control console. 5, Main (or master)
control. 6, Monitoring. 7, Studio
recorder. 8, External recorder.
9, Studio output. 10, Talkback circuit.
In both radio and television outside broadcasts the sound staff
provide communication circuits between operational areas (e.g.
commentary points) and the mobile control room. This often takes
the form of local telephone circuits terminating at special multipurpose boxes that also provide feeds of programme sound and
production talkback. On television outside broadcasts (remotes),
talkback is primarily for the stage manager. In addition, field telephones may be required at remote points. Wireless links are also
widely used.
8
Audio techniques and equipment
In broadcasting, the studio suite (studio and control room) may be
only the first link in a complex chain. In a station with mixed output,
the next link is a continuity suite where the entire output is
assembled. The feed from individual studios is fed in live, and
linked by station identification and continuity announcements. This
announcer may have executive responsibility for the service as a
whole, to intervene if any contribution under-runs, over-runs, breaks
down, or fails to materialize. The announcer also cues recorded
programmes, presents trailers (promotional material) and commercials where appropriate.
In small, specialized stations such as those of local radio, virtually
the whole of the output emerges from a single room ± though a
separate studio may be available for news and discussion programmes or for recording commercials. In this `electronic office'
the presenter controls the microphone and plays in records, recorded
commercials, etc. For this, there should be a range of CD and
MiniDisc players, DAT, reel-to-reel tape decks and cassette players,
together with digital audio editing: sufficient to allow commercials,
promotional trailers, production effects and public service
announcements to be played one after the other (`back-to-back').
Even with a hard-drive play-out system there may be backup from
MiniDiscs, Zip files or other recordings reproduced from `carts' or
stacks. In addition to microphone and reproduction equipment and
telephones (including broadband), the presenter may also have twoway communication with news or traffic cars or helicopters, which
can be arranged to record automatically while on air with other
material.
[1.6] The broadcasting chain
The studio output may go to a
recorder or it may provide an insert
into a programme compiled in
another studio ± or it may be fed
direct to a programme service
continuity suite. Between studio,
continuity and transmitter there may
be other stages: switching centres,
boosting amplifiers and frequency
correction networks on landlines, etc.
1, Studio suite. 2, Recording area or
machine room. 3, Remote studio.
4, Machine room. 5, Other programme
feeds. 6, Continuity announcer.
7, Continuity mixer. 8, Transmitter.
9, Receiver.
Alternatively, much of this may be automated, with output fed from
digital playlists. Audio links are recorded in advance, including
programmed overlaps to previously recorded or musical items. In
this kind of scheduling the studio has a smaller role to play, as more
of the assembly, editing, editorial changes and monitoring of work in
progress are done in an office environment at any convenient time
ahead of transmission ± in particular, without having to wait for a
last-minute run-through to check the result.
In large networks the basic service is put together in either way, but
with greater emphasis on precision timing if the output of the network
continuity appears as a source in the regional or local continuity that
feeds the local radio transmitter. The links between the various centres
may be dedicated landlines or radio links. Broadband (ISDN) telephone lines are increasingly used for some of these feeds.
In television the broadcasting chain is substantially the same as for
radio, except that in the presentation area there may be both a small
television studio and a separate small booth for out-of-vision
announcements. Use of the sound booth means that rehearsals of
more complex material in the main studio do not have to stop for
programme breaks, whether routine or unscheduled.
Chapter 2
The sound medium
Some readers of this book will already have a clear understanding of
what sound is and how it behaves. Others will know very little about
the physics of sound and will not wish to know, their interest being
solely in how to use the medium. These two groups can skip this
section and move directly to the operational techniques, creative
applications and practical problems and their solutions that are the
main theme of this book. Non-technical readers should have little
difficulty in picking up much of what is needed, perhaps with an
occasional glance at the Glossary. Most of the terms used should be
understandable from their context, or are defined as they first appear.
On the other hand, to help make the right choices on questions like
where to place a microphone or what may be done to modify the
signal from it, it will help to know a little about the sound medium
and also how musical sounds are produced by instruments, including
the human voice. In addition, a description of the curious manner in
which our ears make sense of sound may explain some of the very
strange ways in which we measure sound and treat it in our studio
equipment.
The nature of sound
[2.1] Sound wave
A vibrating panel generates waves of
pressure and rarefaction: sound.
A tuning fork is often used for this
demonstration, but in fact tuning
forks do not make much noise. They
slice through the air without moving it
much. A vibrating panel is more
efficient.
Sound is produced by materials that vibrate. If a panel of wood
vibrates, the air next to it is pushed to and fro. If the rate of vibration
is somewhere between tens and tens-of-thousands of excursions per
second, the air has a natural elasticity which we do not find at slower
speeds. Wave your hand backwards and forwards once a second and
the air does little except get out of its way; it does not bounce back.
But if you could wave your hand back and forth a hundred times
every second, the air would behave differently: it would compress as
10
The sound medium
the surface of the hand moves forward and rarefy as it moves back:
the natural elasticity of the air takes over. As the surface moves
forward, each particle of air pushes against the next, so creating a
pressure wave. As the surface moves back, the pressure wave is
replaced by a rarefaction, which travels outward as the air next to it
draws back. And that is followed by another pressure wave, and so on.
[2.2] Waves
1, Sound waves are usually
represented diagrammatically as
transverse waves, as this is
convenient for visualizing them. It also
shows the characteristic regularity of
the simple harmonic motion of a pure
tone. Distance from the sound source
is shown on the horizontal axis, and
displacement of individual air particles
from their median position is shown
on the vertical axis. 2, Because the
particles are moving backwards and
forwards along the line that the sound
travels, the actual positions of layers of
particles are arranged like this. It also
shows the travel of pressure waves.
It is a property of elastic media that a pressure wave passes through
them at a speed that is a characteristic of the particular medium. The
speed of sound in air depends on its temperature as well as the nature
of air. In cool air ± at 15 C (59 F) ± sound travels about 340 metres
in a second (about 1120 feet per second). At 20 C (68 F) this
increases to 344 m/s (1130 ft/s).
This speed is independent of the rate at which the surface generating
the sound moves backward and forward. The example was 100
excursions per second, but it might equally have been 20 or
20 000. This rate at which the pressure waves are produced is called
the frequency, and this is measured in cycles per second, or hertz:
1 Hz is 1 cycle per second.
Let us return to that impossibly energetic hand shaking back and
forth at 100 Hz. It is clear that this is not a perfect sound source:
some of the air slips round the sides as the hand moves in each
direction. To stop this happening with a fluid material like air, the
sound source would have to be much larger. Something the size of a
piano sounding board would be more efficient, losing less round the
edges. But if a hand-sized vibrator moves a great deal faster, the air
simply does not have time to get out of the way. For very high
frequencies, even tiny surfaces are efficient radiators of sound.
In real life, sounds are produced by sources of all shapes and sizes
that may vibrate in complicated ways. In addition, the pressure
waves bounce off hard surfaces. Many sources and all their reflections may combine to create a complex field of intersecting paths.
How are we to describe, let alone reproduce, such a tangled skein of
sound? Fortunately, we do not have to.
[2.3] Frequency and wavelength
All sound waves travel at the same
speed, c, through a given medium, so
frequency, f, is inversely proportional
to wavelength, l.
Consider a single air particle somewhere in the middle of this field.
All the separate pressure waves passing by cause it to move in
various directions, to execute a dance that faithfully describes every
characteristic of all of the sounds passing through. All we need to
know in order to describe the sound at that point is what such a
single particle is doing. The size of this particle does not matter so
long as it is small compared with the separation of the successive
pressure waves that cause the movement. It does not have to be very
small; in fact, we want to measure an average movement that is
sufficiently gross to even out all the random tiny vibrations of the
molecules of air ± just as the diaphragm of the ear does. And
thinking about the ear leads us to another simplification. A single
ear does not bother with all the different directional movements of
the air particle; it simply sums and measures variations in air
pressure. This leads us to the idea that a continuously operating
pressure-measuring device with a diaphragm something like the size
of an eardrum should be a good instrument for describing sound in
The sound medium
11
practical audio-engineering terms. What we have described is the
first requirement of an ideal microphone.
Wavelength
Pressure waves of a sound travel at a fixed speed for a given medium
and conditions, and if we know the frequency of a sound (the
number of waves per second) we can calculate the distance between
matching points on successive waves ± the wavelength. (Wavelength
is often symbolized by the Greek letter lambda, l.)
Taking the speed of sound to be 340 m/s (1120 ft/s), a sound frequency
of 340 Hz (or cycles per second) corresponds to a wavelength
of 1 metre (just over 39 inches). Sometimes it is more convenient
to think in terms of frequency and sometimes wavelength.
Because the speed of sound changes a little with air temperature, any
relationship between wavelength and frequency that does not take
this into account is only approximate. But for most practical purposes, such as calculating the thickness of padding needed to absorb
a range of frequencies or estimating whether a microphone diaphragm is small enough to `see' very high frequencies, an approximate relationship is all we need. Particular frequencies also
correspond to notes on the musical scale:
The correspondence of frequencies to musical notes
Note on piano
Frequency (Hz)
Aiii
Aii
Ai
A
Ai
Aii
Aiii
Aiv
14 080
3 520
1 760
880
440
220
110
55
27.5
Wavelength
2.5 cm
9.5 cm
19 cm
38 cm
75 cm
1.5 m
3m
6m
12 m
1 in
3.75 in
7.5 in
1 ft 3 in
2 ft 6 in
5 ft
10 ft
20 ft
40 ft
Here the speed of sound is taken to be 335 m/s (1100 ft/s), as it may be in cool air.
Note that the length conversions and shorter wavelengths are approximate only.
[2.4] Piano keyboard
Note and frequency of fundamental,
octave by octave.
The first thing to notice from this is the great range in physical size
these different frequencies represent. Two octaves above the top A
of the piano, the wavelength is approximately 25 mm (1 in). As it
happens, this is close to the upper limit of human hearing; and the
corresponding size is found in the dimensions of high-quality microphones: they must be small enough to sample the peaks and troughs
of this wave. At the lower end of the scale we have wavelengths of
12 m (40 ft). The soundboard of a piano can generate such a sound,
though not so efficiently as some physically much larger air cylinders in a pipe organ. Below the 20 Hz lower limit of human hearing,
the deepest note of an organ may be 16 Hz. Only its overtones are
audible as sound; the fundamental is sensed as vibration.
12
The sound medium
The second point to notice, and the reason for the great range of
sizes, is that for each equal musical interval, a rise of one octave, the
frequency doubles (and the wavelength is halved).
[2.5] Velocity of sound in air
The velocity of sound increases with
temperature.
Much less obvious is the fact that as the pitch of a sound depends on
the size of the object making it, and because that size remains nearly
constant as temperature increases, then because the velocity of sound
increases with temperature, the pitch goes up. Metal strings expand,
so go a little flat, while wind instruments go sharp. For strings, the
remedy is simple: they can easily be tuned, but the vibrating column
of air in most wind instruments cannot be changed so easily. A flute,
for example, sharpens by a semitone as the air temperature goes up
by 15 C. (As the metal of the instrument itself warms up and so
lengthens by a tiny amount, this means that it also flattens a little, but
this is not by enough to make much difference.) To compensate for
weather, hall and instrument temperature changes, the head joint of
the flute can be eased in or out a little ± in an orchestra, matching its
pitch to that of the traditionally less adaptable oboe.
Waves and phase
Still considering a pure, single-frequency tone, we can define a few
other variables and see how they are related to the pressure wave.
[2.6] Waveform relationships
1, Pressure wave. 2, Displacement
wave. 3, Pressure gradient. 4, Particle
velocity.
[2.7] Adding sound pressures
At any point the sound pressure is the
sum of the pressures due to all
waves passing through that point.
If the simple waves A and B are
summed this results in the pressure
given by the curve C. 1, Here there is
partial cancellation. 2, Here the peaks
reinforce. The resultant curve is more
complex than the originals.
First we have particle velocity, the rate of movement of individual
air particles. As this is proportional to air pressure, the waveforms
for pressure and velocity are similar in shape. But not only that:
where one has a peak the other has a peak; where there is no excess
of pressure, there is no velocity, and so on. The two waveforms are
said to be in phase with each other.
The next variable is pressure gradient, the rate at which pressure
changes with distance along the wave. Plainly, this must have a
waveform derived from that of pressure, but where the pressure is at
a peak (maximum or minimum) its rate of change is zero. Again it
turns out that for a pure tone (which has the shape of what in
mathematics is called a `sine wave') the shape of the wave is the
same, but is displaced a quarter of a wavelength to one side. The two
waveforms are said to be a quarter of a wavelength out of phase.
Another variable, again with a similar waveform, is that for particle
displacement. This is a graph showing how far a particle of air has
moved to one side or the other of its equilibrium position and how
this changes in time. (This is, for sound, the equivalent of the wave
we see moving over water.) Displacement is proportional to pressure
gradient, and is therefore in phase with it. It is directly related
to amplitude.
These terms are all encountered in the theory of microphones. Some
(such as the moving-coil and ribbon types that will be described in
Chapter 5) are devices that measure the velocity of the air particles
next to them. Others (condenser or crystal types) measure the
The sound medium
13
amplitude of the sound wave. A curious effect of this is that, while
the information they transmit is virtually identical, the electrical
waveforms that carry them are slightly different. More will, however, be heard of pressure and pressure-gradient modes of operation,
as these characteristics of microphones lead to important differences
in the way they can be used.
Because of these differences, the phase of the signal as it finally
appears in the microphone output is not always the same as that of
the sound pressure measured by the ear. Fortunately the ear is not
usually interested in phase, so it does not generally matter whether a
microphone measures the air pressure or pressure gradient. The only
problem that might occur is when two signals very similar in form
but different in phase are combined. In an extreme case, if two equal
pure tones that are exactly half a wavelength out of phase are added,
the output is zero. Normally, however, sound patterns are so complex that microphones of completely different types that are sufficiently far apart may be used in the same studio, and their outputs
mixed, without much danger of differences of phase having any
noticeable effect.
So far we have considered simple tones. If several are present
together, a single particle of air can still be in only one place at a
time: the displacements are added together. If many waves are
superimposed, the particle performs the complex dance that is the
result of adding all of the wave patterns. Not all of this adding
together results in increased pressure. At a point where a pressure
wave is superimposed on a rarefaction, the two ± opposite in phase at
that point ± tend to cancel each other out.
Waves and obstacles
Close to most sources sound radiates outward in all directions that
are open to it, in the simplest configuration as a spherical wave.
Close to a large, flat sound source, sound radiates initially as a plane
wave. In addition, a spherical wave gradually loses curvature as it
gets further from the source, becoming more and more like a plane
wave. Spherical and plane waves behave differently when they
encounter microphones (see Chapter 5) or meet obstacles in a studio.
[2.8] Reflected sound
1, The angle of incidence is the same
as the angle of reflected sound, but
there is some loss as sound is
absorbed. 2, At a 90 corner, sound
is reflected in the direction it came
from. 3, A convex surface disperses
and so diffuses sound. 4, A concave
surface concentrates sound; a
parabolic reflector focuses it.
Sound waves reflect from hard surfaces in the same way that light
reflects from shiny objects, and the familiar optical rules apply.
Sound approaching a flat surface at an angle is reflected at the same
angle, apart from some proportion that is absorbed. This is one of the
basic features of studio acoustics (see Chapter 4). Double reflections,
in a right-angled corner, send it back in the direction it came from.
If sound encounters a surface that bows outward, it will disperse.
This is one way of creating sound diffusion; others are to place
uneven surfaces or a variety of objects in its path (diffusion is
another important feature of studio acoustics and also of reverberation). If sound encounters a concave surface it is focused and
14
The sound medium
concentrated, making it louder without the need for added power. A
plane wave going into a parabolic reflector is focused to a point (so a
microphone at the focus can pick up extremely faint, distant sounds).
Diffraction is another effect that is like those that can be produced
with light. If a surface has a hole in it that is small compared to the
wavelength of sound approaching it, on the other side the hole
radiates sound like a new point-source. This radiation is so faint that
it is of no practical importance in studio acoustics ± but this is also
what happens inside some small microphone capsules.
Of greater practical importance is diffraction around an object in the
path of a wave. If the wavelength is long compared with the width of
the object, the wave is virtually undisturbed: it is as though the wave
does not `see' the object (but to reach the back of the object it has to
travel further than if it could go right through it). If the obstruction is
wider than the wave is long, it casts a shadow.
Energy, intensity and resonance
There are several more concepts that we need before going on to
consider the mechanics of music.
[2.9] Diffraction
1, An obstacle that is smaller than
a wavelength leaves the wave
undisturbed. 2, An obstacle that is
broader than a wavelength casts a
shadow. 3, A small (or narrow) gap
acts as a new source. 4, A broad gap
allows a wave to pass undisturbed,
with shadows on either flank.
[2.10] Sound intensity
This is the energy passing
through unit area per second.
For a spherical wave (i.e. a wave
from a point source) the
intensity dies away very rapidly
at first. The power of the sound
source S is the total energy
radiated in all directions.
First, the energy of a sound source. This depends on the amplitude of
vibration: the broader the swing, the more power (energy output per
second) it can produce. The sound intensity at any point is then
measured as the acoustic energy passing through unit area per
second. But to convert the energy of a sound source into acoustic
energy in the air, we have to ensure that the sound source is properly
coupled to the air, so that its own vibrations are causing the air to
vibrate with it. Objects that are small (or slender) compared with the
wavelength in air associated with their frequency of vibration (e.g.
tuning forks and violin strings) are able to slice through the air
without giving up much of their energy to it: the air simply slips
around the sides of the prong or the string.
If a tuning fork is struck and then suspended loosely it goes on
vibrating quietly for a long time. If, however, its foot is placed on a
panel of wood, the panel is forced into vibration in sympathy and
can transmit vibration to the air. The amplitude of vibration of
the tuning fork then goes down as its energy is lost (indirectly) to
the air.
The sound medium
15
[2.11] Tuning fork
1 and 2, Each fork vibrates at a
specific natural frequency, but
held in free air radiates little sound.
3, Placed against a wooden panel,
the vibrations of the tuning fork are
coupled to the air more efficiently
and the fork is heard clearly.
4, Placed on a box having a cavity
with a natural resonance of the
same frequency as the tuning fork,
the sound radiates powerfully.
If the panel's natural frequency of vibration is similar to that of the
tuning fork, the energy can be transferred and radiated much faster.
In fact, wooden panels fixed at the edges may not appear to have
clear, specific frequencies at which they resonate: if you tap one it
may make a rather dull sound, though perhaps with some definable
note that seems to predominate. In the violin, the wooden panels are
very irregular, which ensures that over a whole range of frequencies
no single one is emphasized at the expense of others. The string is
bowed but transmits little of its energy directly to the air. Instead, the
energy is transferred through the bridge that supports the string to
the wooden radiating belly of the instrument.
[2.12] Complex waveforms
Pure tones (1, 2 and 3) may be added
together to produce a composite
waveform (4). In this case the first
two overtones (the second and third
harmonics) are added to
a fundamental. If an infinite series
of progressively higher harmonics
(with the appropriate amplitudes and
phases) were added, the resultant
waveform would be a `saw-tooth'.
When the partials are related
harmonically, the ear hears
a composite sound having the pitch
of the fundamental and a particular
sound quality due to the harmonics.
But when the pitches are unrelated or
the sources are spatially separated,
the ear can generally distinguish them
readily: the ear continuously analyses
sound into its components.
At this stage an interesting point arises: the radiating panels are not
only smaller (in length) than the strings themselves, but also considerably smaller than some of the wavelengths (in air) that the
strings are capable of producing. The panels might respond well
enough to the lowest tones produced, but they are not big enough to
radiate low frequencies efficiently. This leads us to consider several
further characteristics of music and musical instruments.
Overtones, harmonics and
formants
Overtones are the additional higher frequencies that are produced
along with the fundamental when something like a violin string or
the air in an organ pipe is made to vibrate.
On a string the overtones are all harmonics, exact multiples of the
fundamental, the lowest tone produced. As in most musical instruments, it is this lowest tone that defines its pitch. If the string were
bowed in the middle, the fundamental and odd harmonics would be
16
The sound medium
[2.13] The open strings of the
violin: intensities of harmonics
The structure of the resonator
ensures a very wide spread, but is
unable to reinforce the fundamental
of the low G, for which it is physically
too small. Note that there is a
difference in quality between the
lowest string (very rich in harmonics)
and the highest (relatively thin in tone
colour).
emphasized, as these all have maximum amplitude in the middle of the
string; and the even harmonics (which have a node in the middle of
the string) would be lacking. But the string is, in fact, bowed near one
end so that a good range of both odd and even harmonics is excited.
There is one special point here: if the string is bowed at approximately one-seventh of its length it will not produce the seventh
harmonic, as this would have a node, a point where the string is
not displaced, at that point. So, for the most harmonious quality, that
is a good place to bow at, because as it happens the seventh harmonic is the first that is not musically related to the rest. The sixth
and eighth harmonics are both full members of the same musical family,
while higher harmonics (whether family members or not) are produced in such rich profusion that their main role is to add a dense
tonal texture high above the fundamental. These are all characteristics of what we call `string quality', in any stringed instrument.
[2.14] The first eight string
harmonics
The fundamental is the first
harmonic: the first overtone is the
second harmonic, and so on. They
are all notes in the key of the
fundamental, except for the
seventh harmonic, which is not a
recognized note on the musical
scale (it lies between G and G#). If
the violinist bows at the node of this
harmonic (at one-seventh of the
length of the vibrating string) the
dissonance is not excited. The
eleventh harmonic (another
dissonant tone) is excited, but is
in a region where the tones are
beginning to cluster together to
give a brilliant effect.
The character of the instrument itself ± violin, viola, cello or bass ±
is defined by the qualities of its resonator, and most particularly by
its size. The shape and size of the resonator superimposes on the
string quality its own special formant characteristics. Some frequencies, or ranges of frequencies, are always emphasized; others are
always discriminated against. Formants are very important:
obviously, in music they are a virtue, but in audio equipment the
same phenomenon might be called `an uneven frequency response'
and would be regarded as a vice ± unless it were known, calculated
and introduced deliberately with creative intent.
The sound medium
17
[2.15] Bell overtones
Left: Profiles and tones produced
by 1, a minor-third bell and 2, a
major-third bell. Right: The relative
volumes and rates of decay of the
tones present in the sound of a
minor-third bell.
On some instruments, such as bells, getting the principal overtones to
have any musical relationship to each other is a major manufacturing
problem (in fact, the lowest note of a bell is not even distinguished
by the name of fundamental). In drums, the fundamental is powerful
and the overtones add richness without harmonic quality, while from
triangles or cymbals there is such a profusion of high-pitched tones
that the sound blends reasonably well with almost anything.
An important feature of any musical instrument is that its vibrating
surfaces or columns of air do not radiate equally efficiently in all
directions. Different frequencies, too, may radiate more or less
powerfully in different directions, so that the mixture of tones
changes with angle.
There are many other essential qualities of musical instruments.
These may be associated with the method of exciting the resonance
(bowing, blowing, plucking or banging); or with qualities of the note
itself, such as the way it starts (the attack, which creates an irregular
transient), is sustained and then released, changing in volume as the
note progresses (to create its envelope).
[2.16] Vibration of a drumskin
(a stretched circular membrane,
clamped at the edges) The suffixes
refer to the number of radial and
circular nodes (there is always a
circular node at the edge of the skin).
The overtones are not harmonically
related. If f01 ˆ 100 Hz, the other
modes of vibration shown here are:
f02 ˆ 230, f03 ˆ 360, f11 ˆ 159,
f12 ˆ 292, f21 ˆ 214 Hz.
18
The sound medium
[2.17] Wind instrumental formants
1, Oboe. 2, Horn. 3, Trombone.
4, Trumpet. The formants, imposed
by the structural dimensions of
parts of the instruments, may be
broad or narrow. They provide an
essential and recognizable
component of each instrument's
musical character.
Air resonance
Air may have dimensional resonances very much like those of a
string of a violin, except that whereas the violin string has transverse
waves, those in air, being composed of compressions and rarefactions, are longitudinal waves. Radiated sound moves through air in
the form of progressive waves, but dimensional resonances stand
still: they form stationary or standing waves.
These stationary waves can again be represented diagrammatically
as transverse waves. The waveform chosen is usually that for displacement amplitude. This makes intuitive sense: at nodes (e.g. at
solid parallel walls if the resonance is formed in a room) there is no
air-particle movement; while at antinodes (e.g. halfway between the
walls) there is maximum movement of the air swinging regularly
back and forth along a line in the room. Harmonics may also be
present.
Standing waves are caused when any wave strikes a reflecting surface at a right angle and travels back along the same path. A double
reflection in a right-angled corner produces the same effect,
although reduced by the greater loss in two reflections.
The air in a narrower space (such as an organ pipe) can also be made to
resonate. Where an end is closed, sound will reflect, as at any other solid
surface. But if the pipe is open at the end, resonance can still occur,
because in a tube that is narrow compared with the length of the wave
[2.18] Stationary wave
Stationary waves are formed from
progressive waves moving in opposite
directions.
The sound medium
19
[2.19] Vibration of air columns
Left: For an open pipe, the
fundamental frequency, f, is twice
the length of the pipe. There is a node
N at the centre of the pipe and
antinodes at the open ends. The
second and third harmonics are
simple multiples of the fundamental:
f2 ˆ 2f1, f3 ˆ 3f1, etc. Left below: The
third harmonic, half a cycle later. Air
particles at the antinodes are now
moving in the opposite direction, but
the air at the nodes remains
stationary. Right: A pipe closed at one
end (i.e. stopped). The wavelength of
the fundamental is four times the
length of the pipe. The first overtone
is the third harmonic; only the odd
harmonics are present in the sound
from a stopped pipe: f, 3f, 5f, etc. In
both types of pipe most of the sound
is reflected back from an open end
for all frequencies where the aperture
is small compared with the
wavelength.
the sound has difficulty in radiating to the outside air. The energy has to
go somewhere: in practice, a pressure wave reflects back into the tube as
a rarefaction, and vice versa. In this case the fundamental (twice the
length of the pipe) and all of its harmonics may be formed. If one end is
open and the other closed, only the fundamental (the wavelength of
which is now four times the length of the tube) and its odd harmonics
are formed. The tone quality is therefore very different from that of a
fully enclosed space, or again, from that of a pipe that is open at both
ends.
Wind instruments in the orchestra form their sounds in the same
way: the length of the column may be varied continuously (as in a
slide trombone), by discrete jumps (e.g. trumpet and horn), or by
opening or closing holes along the length of the body (e.g. flute,
clarinet or saxophone). Formants depend on the shape of the body
and of the bell at the open end ± though for many of the notes in
instruments with finger holes the bell makes little difference, as most
of the sound is radiated from the open holes.
[2.20] Helmholtz or cavity resonator
Air inside the cavity vibrates like a
weight hanging on a spring.
Another important way in which air resonance may govern pitch is
where a volume of air is almost entirely enclosed and is connected to
the outside through a neck. Such a device is called a Helmholtz or
cavity resonator. It produces a sound with a single, distinct frequency,
20
The sound medium
an example being the note obtained by blowing across the mouth of a
bottle. In a violin the cavity resonance falls within the useful range of
the instrument and produces a `wolf-tone' that the violinist has to treat
with care, bowing it a great deal more gently than other notes.
The voice
The human voice is a great deal more versatile than any musical
instrument. This is because the cavities of the mouth, nose and throat
impose such dramatically variable formant characteristics on the
sounds modulated by the vocal cords. It is as though a violin had five
resonators, several of which were continuously changing in size, and
one (the equivalent of the mouth) so drastically that it completely
changes the character of the sound from moment to moment.
These formant characteristics, based on cavity resonance, are responsible for vowel sounds and are the main vehicle for the intelligibility of speech. Indeed, sometimes they are used almost on their
own: an example is a robot speech effect created by extracting
formants from a human voice (or simulating them by computer)
then using them to modulate some simple continuous sound, the
nature of which is chosen to suggest the `personality' of the machine.
[2.21] Vocal cavities
1, Lungs. 2, Nose. 3, Mouth. This is
the most readily flexible cavity, and is
used to form vowel sounds.
4, Pharynx, above the vocal cords.
5, Sinuses. Together, these cavities
produce the formants characteristic
of the human voice, emphasizing
certain frequency bands at the
expense of others.
[2.22] Human speech analysed to
show formant ranges
1, Resonance bands. 2, Pause before
plosive. 3, Unvoiced speech. 4, Voiced
speech. These formants arise from
resonances in nose, mouth and throat
cavities. They are unrelated to the
fundamental and harmonics, which
are shown in the second analysis of
the word `see'. 5, The vocal
harmonics fall as the voice is dropped
at the end of the sentence. At the
same time, the resonance regions
rise, as the vocal cavities are made
smaller for the `ee' sound.
A number of other devices are used in speech: these include sibilants
and stops of various types which, together with formant resonances,
provide all that is needed for high intelligibility. A whisper, in which
the vocal cords are not used, is clear and understandable; in a stage
whisper, intelligibility carries well despite lack of vocal power.
The vibrations produced by the vocal cords add volume, further
character and the ability to produce song. For normal speech the
fundamental may vary over a range of about 12 tones and is centred
somewhere near 145 Hz for a man's voice and 230 Hz for a
woman's. The result of this is that, as the formant regions differ
little, the female voice has less harmonics in the regions of stronger
resonance; so a woman may have a deep-toned voice, but its quality
is likely to be thinner (or purer) than a man's. For song, the fundamental range of most voices is about two octaves, although, exceptionally, it can be much greater.
The sound medium
21
The human ear
One of the functions of the outer ear is to act as a direction finder: we
will return to this when we look at how stereo works. For the
moment, just follow the sound in to the part that makes sense of it.
[2.23] The ear
1, Sound enters via the outer ear and
the auditory canal. This channel has a
resonance peak in the region of
3±6 kHz. At the end of the auditory
canal is the eardrum, which vibrates in
sympathy with the sound but is not
well adapted to following large lowfrequency excursions. 2, The middle
ear contains air, the pressure in which
is equalized through the Eustachian
tube, a channel to the pharynx,
behind the tonsils. With air on both
sides, the eardrum vibrates to feed
the sound mechanically across the
middle ear along three small bones,
ossicles, that form an impedancematching lever mechanism: this
converts the acoustic energy of the
air to a form suitable for transmission
through the fluid of the inner ear.
Small muscles attached to the
ossicles tighten up in response to
loud noise: this aural reflex
temporarily reduces the effect of any
immediate repetition. 3, The sound is
pumped via a membrane called the
oval window into the tiny delicate
channels of the inner ear, the cochlea.
4, The cochlea is shaped like a shell,
with two channels along its length
that get narrower until they join at the
far end. Distributed along a vibrating
membrane between the channels are
hair-like cells that may respond to
particular frequencies: when any
`hair' is bent a nerve impulse is fired.
The further it lies along the canal, the
lower the frequency recorded.
5, A bundle of 4000 nerve fibres carries
the information to the brain, where
it is decoded. 6, The pressure wave
returns via another membrane, the
round window, to dissipate back in
the middle ear.
The part of the ear that senses sound is a tiny spiral structure called
the cochlea. It is a structure that gets narrower to one end of the coil,
like the shell of a snail. But unlike the shell, it is divided lengthways
into two galleries, which join only at the narrowest inner end. The
entire structure is filled with fluid to which vibrations may be
transmitted through a thin membrane or diaphragm called the oval
window. The acoustic pressures may then travel down one side of
the dividing partition (the basilar membrane) and return up the other,
to be lost at a further thin diaphragm, the round window.
All along one side of the basilar membrane are `hairs' (actually
elongated cells) that respond to movements in the surrounding fluid.
Each of these hairs standing in the fluid acts as a resonator designed
to respond to a single frequency (or rather, to a very sharply tuned
narrow band). The hairs therefore sense sound not as an air particle
sees it, as a single, continuous, very complex movement, but as
a very large number of individual frequencies. The hairs are so
arranged as to give roughly equal importance to equal musical
intervals in the middle and upper middle ranges, but the separation
is greater at very low frequencies: this is why it is more difficult
to distinguish between very low pure tones. So (apart from indicating
this lack of interest in low notes) a study of the mechanism of
hearing confirms that we were right to be concerned with musical
intervals that are calculated by their ratios rather than on a linear
scale.
Growth of this sort is called exponential: it doubles at equal intervals, and a simple graph of its progress rapidly runs out of paper
unless the initial stages are so compressed as to lose all detail. To
tame such growth, mathematicians use a logarithmic scale, one in
which each doubling is allotted equal space on the graph. The
frequencies corresponding to the octaves on a piano progress in
the ratios 1:2:4:8:16:32:64:128. The logarithms (to base 2) of
these numbers are in a much simpler progression ± 0, 1, 2, 3, 4, 5,
6, 7 ± and these equal intervals are exactly how we perceive the
musical scale.
Frequency is not the only thing that the ear measures in this way:
changes of sound volume follow the same peculiar rule.
Sound volume and the ear
The hairs on the basilar membrane vibrate in different degrees,
depending on the loudness of the original sound. Here again they
measure not by equal increases in loudness but by ratios of loudness.
22
The sound medium
Now we have a choice (which is again because of the way in which
ratios behave, and can be manipulated mathematically by logarithms
± a point that those who really hated school mathematics do not
really need to understand). The choice is whether we describe
changes in loudness in terms of sound intensity or in terms of power.
As it happens, for most purposes, the choice has already been made
for us ± because engineers have decided to describe changes in the
volume of an audio signal passing through electrical equipment in
terms of its power. (There is a slight difference in the way power and
sound intensity are calculated.)
In our hearing the practical effect of the way this works is that each
doubling in power sounds roughly as loud again as the last. It is
therefore convenient once again to use a logarithmic scale, and the
measure that is employed is the decibel, or dB. This is based on a
unit called the bel (from the name of the pioneer audio engineer,
Alexander Graham Bell), which corresponds to a decimal decrease
or increase. As it turns out, the bel is inconveniently large, because
the ear can detect much smaller changes. So an interval a tenth of the
size ± the decibel ± has been adopted.
[2.24] Exponential and logarithmic
scales
1, Exponential growth rapidly goes off
the top of the scale (in this case a long
way off the top of the page). 2, The
corresponding logarithmic growth
stays within reasonable bounds.
One decibel represents a ratio of about 1.26:1 (because 1.26 is
approximately the tenth root of 10). This is just about as small a
difference in power as the human ear can detect in the best possible
circumstances.
As it happens, a power ratio of 3 dB is almost exactly 2:1. This is
convenient to remember, because if we double up a sound source we
double its power. So if we have one soprano singing with gusto and
another joins in, equally loudly, the sound level will go up 3 dB (not
all that much more than the minimum detectable by the human ear).
But to raise the level by another 3 dB, two more sopranos are
needed. Another four are needed for the next 3 dB ± and so on.
Before we get very far, we have to add sopranos at a rate of 64 or
128 a time to get any appreciable change of volume out of them. To
increase intensity by increasing numbers soon becomes extremely
expensive, so if volume is the main thing you want it is better to
choose a more powerful source to begin with ± such as a pipe organ,
a trombone, or a bass drum!
Loudness, frequency and
hearing
[2.25] Sound decay
1, Logarithmic decay, compared with
examples of possible real sounds.
2, Attack, followed by sustained note
then 3, release and rapid decay of
reverberation. 4, Attack followed by
rapid decay then sustained
reverberation.
The ear does not measure the volume of all sounds by the same
standards. Although for any particular frequency the changes in
volume are heard more or less logarithmically, the ear is more
sensitive to changes of volume in the middle and upper-middle
frequencies than in the bass.
The range of hearing is about 20±20 000 Hz for a young person, but
the upper limit of good hearing falls with age to 15 000 or 10 000 Hz
The sound medium
[2.26] Equal loudness contours
Left: The classical curves of
Fletcher and Munson, published
in 1933. Intensity (dB) equals
loudness in phons at 1 kHz.
Right: The curves measured by
Robinson and Dadson take age
into account and show marked
peaks and dips in young hearing.
The unbroken lines are for age
20; the broken lines show typical
loss of high-frequency hearing at
age 60. The lowest line in each
set represents the threshold of
hearing.
dB
140
130
Bass drum, at 3 cm
Loud singer, near mouth
120
Tom-toms, at 3 cm
Congas, at 2 cm
Electric guitar
loudspeaker, 30 cm
110
Cowbell, at 10 cm
100
Loud singer, at 15 cm
90
Saxophone or trombone
played softly, at 40 cm
80
Acoustic guitar with
plectrum, 40 cm
70
Piano played softly, 1 m
Acoustic guitar,
fingers, 40 cm
60
50
Whisper at 10 cm,
quit conversation, 1 m
40
Ambient noise in typical
room in a city
30
20
Ambient noise in quiet
recording studio
10
Inside sound-proofed room
0
Threshold of hearing
[2.27] Sound pressure levels
Examples of sounds over the wide
range that may reach microphone or
ear.
23
(and for many people this fall-off is much steeper). Sensitivity is
greatest at 1000 Hz and above, and best at 2000±3000 Hz: the auditory canal between the outer ear and the eardrum helps this by
having a broad resonance in the region of 2000±6000 Hz.
Obviously loudness (a subjective quality) and the measurable
volume of a sound are not the same thing, but for convenience they are
regarded as being the same at 1000 Hz. Perceived loudness, originally described in units called phons, can then be calculated from
actual sound volume by using a standard set of curves representing
typical good human hearing. From these curves it is also clear that
people with hearing loss get a double benefit if sound equipment is
turned up louder: at progressively higher volumes, more of the high
(and low) frequencies are heard, which profoundly affects the
intelligibility of speech.
Figures for such things as levels of noise are also weighted to take
hearing into account. This may refer to electrical as well as acoustic
noise. For example, some types of microphone produce more noise
in the low-frequency range: this is of less importance than if the
noise were spread evenly throughout the entire audio range. Today,
engineers use the unit `dB(A)' rather than the phon.
The lower limit is called the threshold of hearing. It is convenient to
regard the lower limit of good human hearing at 1000 Hz as zero on
the decibel scale. There is no natural zero: absolute silence would be
minus infinity decibels on any practical scale. (The zero chosen
corresponds to an acoustic pressure of 2 10 5 pascals, or Pa for
short. One pascal is a force of one newton per square metre.)
The upper practical limit is set by the level at which sound begins to
be physically felt ± a threshold of feeling at around 120 dB. In some
experiments, 50% of the subjects reported feeling sound at 118 dB
and that it became painful at 140 dB ± a volume well above the level
at which repeated or lengthy exposures can damage hearing.
When we hear two loud tones together, and they are separated by
less than about 50 Hz, we also hear a `beat', an oscillation in volume
at a frequency equivalent to the difference between the two original
tones. If one of the tones is variable, this effect can be used to
synchronize pitch. If the difference between the two loud tones
increases to more than 50 Hz, the beat changes to an additional
difference tone. Increase the separation further, beyond a semitone,
and this spurious combination tone gradually disappears.
24
The sound medium
Another effect we may experience is called masking. When the
sound from two neighbouring instruments contains similar tones at
different volumes, the louder tone may completely mask the softer
one. This may change the character of the quieter sound ± requiring
treatment, either to strengthen the masked component (see Chapter 8)
or to increase the apparent spatial separation. Note that loud harmonics in the louder sound can also cause masking.
Both loudness and frequency range are matters of vital importance in
the design of loudspeakers, though for those working in the sound
medium this is mainly restricted to monitoring loudspeakers. But the
frequency range that human ears can respond to has important
applications for most other components of the chain, and especially
to microphones.
Frequency range is particularly important to broadcasting transmission standards. An individual may be happy to spend money on
equipment with a frequency response that is substantially level from
20 to 20 000 Hz, knowing this goes beyond the limits most listeners
will require, but for a broadcasting organization this may be too
high. For most people hearing is at best marginal at 15 000 Hz, and
many television receivers produce a whistle only a little beyond this.
So it is wasteful to go above 15 000 Hz ± particularly as the amount
of information to be transmitted does not observe the same logarithmic law that the ear does. The octave between 10 000 and 20 000 Hz
is by far the most `expensive' in this respect.
[2.28] Eardrum and microphone
diaphragm
The shortest wavelength that young
ears can hear defines the size of
diaphragm that is needed for a
high-quality pressure microphone.
Most people seem to expect that recording equipment (and many
other items in an audio chain) should respond to the full range from
20 Hz to 20 kHz (although few loudspeakers or microphones can do
this). Some people experience hums below 20 Hz, and find them
deeply unpleasant; plainly, these frequencies should be avoided. The
most controversial range is 20 Hz to 100 kHz: some people report
that whether they can `hear' that or not, its presence adds to their
listening pleasure. But a practical range for many broadcasting
transmitters is 30±15 000 Hz.
Chapter 3
Stereo
The object of stereo is to lay before (or around) the listener a lifelike
array of sound ± both direct sound and reflections from the walls of
the original studio ± to recreate some real or simulated layout in
space.
In its simplest form this sound stage is created by two loudspeakers,
each of which has a separate signal fed to it. Sound that is reproduced by a single loudspeaker appears to come directly from it. If,
however, the sound is split evenly and comes equally from the two
speakers it seems as though it is located halfway between them.
Other mixtures being fed to the two loudspeakers simulate other
directions on the sound stage. It is usual to refer to the left-hand
loudspeaker as having the `A' signal and that on the right as having
the `B' signal. (Some manufacturers called them `X' and `Y'.)
A serious disadvantage of mono is that reverberation is superimposed on the same point source of sound, reducing the clarity of
music and intelligibility of speech. In stereo, the reverberation is
spread over the full width of the sound stage ± a vast improvement,
although in two-channel stereo it still does not surround the listener.
[3.1] Hearing stereo
The position of apparent sources is
perceived as the brain processes
time, amplitude and phase
differences between signals received
by the left and right ears, and also
(at high frequencies) by the effects of
shadowing due to the head. For any
particular frequency the phase at
each ear is given by the sum of the
sounds arriving from the two
loudspeakers.
Two loudspeakers
A stereo pair of loudspeakers must work together in every sense. The
minimum requirement is that if the same signal is fed to both, the
cones of both loudspeakers move forward at the same time; in other
words, the loudspeakers must be wired up to move in phase with
each other. If they move in opposite senses an effect of sorts is
obtained, but it is not that intended. For example, a sound that should
be in the centre of the sound stage, the most important place, is not
clearly located in the middle, but has an indeterminate position.
26
Stereo
[3.2] Stereo loudspeakers: spacing
With the closer loudspeakers (left), the
spread of the stereo image is reduced,
but there is a broad area of the
listening room over which reasonable
stereo is heard. With broader spacing
(right), the stereo image has more
spread, but for the benefit of a
relatively narrow region in which it can
be heard at its best. For listeners
off the centre line C, the growing
difference in distance rapidly `kills'
the stereo, due to the Haas effect.
Listening position 1 (with the narrower
spacing of speakers) has the same
loss of stereo effect as position 2
(with the broader spacing).
A second requirement is that the loudspeakers be matched in volume
and frequency response. If the volume from one is set to be greater
than that from the other, the centre is displaced and the sound stage
distorted. If the difference reaches 20 dB, the centre is displaced to
the extreme edge of the sound stage. For most listeners the exact
matching of frequency responses is less important. Probably the
most important effect of a mismatch is that it could draw attention
to the loudspeakers themselves and diminish the illusion of reality.
In the studio, anyone who must listen critically to sound needs to
work in fully standardized conditions. This generally means sitting
on the centre line of a pair of matched high-quality loudspeakers
placed about 2.5 m (8 ft) apart. It is also recommended that a home
listener should have loudspeakers 2±3 m (between 6 and 10 ft) apart,
depending on the size of the room. With less separation the listener
should still hear a stereo effect, but the width of the sound stage is
reduced. On the other hand, the speakers should not be too far apart.
For normal home listening, if one person is on the centre line, others
are likely to be 60 cm (2 ft) or more to the side, and the further the
loudspeakers are apart, the greater the loss of stereophonic effect to
the outer members of the group. If a listener is as little as 30 cm (1 ft)
closer to one loudspeaker, phase effects that distort the sound stage
are already marked (see later), and as the difference in distance grows,
the nearer source progressively appears to capture the whole signal.
[3.3] Haas effect
For a listener away from the centre
line C, the difference in distances
means that the sound from the
farther loudspeaker arrives at the
listener's ears later than that from the
nearer one. The delay is 1 ms for
every 34 cm (13.4 in) difference in
path length. For identical signals
coming from different directions, a
difference of only a few milliseconds
makes the whole sound appear to
come from the direction of the first
component to arrive. Increasing the
volume of the second signal can
reverse the effect: the curve shows
the increase in volume required to
recentre the image.
This second distortion of the sound stage ± indeed, its near collapse ±
is induced by the Haas effect. This is one of the ways in which the
ear can tell the source of a sound: it can distinguish direct from
reflected sound that is delayed by as little as a few thousandths of a
second. This happens even in a small room: in 1 ms (one millisecond), sound travels just over 34 cm (131¤2 in). Sit much closer to
one of the loudspeakers and the Haas effect will fool you.
The Haas effect is at its strongest for delays of 5±30 ms, which are
produced by a difference in path lengths of about 1.7±10 m (roughly
6±33 ft). The way ears respond to this effect diminishes a little with
increasing delay, but remains effective until eventually the reflection
of a staccato sound can be distinguished as a separate echo.
What happens at the lower end of this scale of delays and their
corresponding distances is even more important for our perception
of stereo. As the difference in distance from the two loudspeakers
grows from zero to about 1.5 m (5 ft), the Haas effect grows rapidly
and continuously to approach its maximum, at which it would take
Stereo
27
an increase of 10 dB in volume from the farther loudspeaker for it to
begin to recapture the source. Plainly, an optimum stereo effect is
achieved only in a narrow wedge-shaped region along the centre line,
and this is narrower still if the loudspeakers are widely separated.
Between the reduced spatial effect of narrow separation and the
reduced optimum listening area created by a broad layout, it
may seem surprising that stereo works at all. The fact is, most
casual listeners happily accept a sound picture that is spatially
severely distorted. This is their prerogative: the professional should
be aware that listeners often prefer their own seating plan to a
clinical ideal.
Reduced loudspeaker separation
[3.4] Early stereo surround for
70 mm film
A, To cover its full width evenly,
1950s 70 mm film had six magnetic
tracks. 1±5, Five loudspeakers,
spread across the picture. 6, One
track fed an array of loudspeakers
around the auditorium. B, In a variant
of this the six tracks fed: 7, Left.
8, Centre. 9, Right. 10, Low-frequency
loudspeakers. 11, Left surround.
12, Right surround.
Many people accept a reduced loudspeaker separation in their own
homes, often from portable or compact audio equipment that has
loudspeakers built in or attached on either side. Television or video
stereo is also usually heard through loudspeakers within the box
itself. This reduces the perceived sound stage to a fraction of that
available to the music balancer. For television, it may match the size
of the visual image, but limits any suggestion of off-screen sound
sources.
Fortunately there are compensations. One is that the Haas effect is
reduced, permitting off-centre-line viewing, as is necessary for
family groups. There is still some separation of point sources
and, equally important, reverberation is spread away from the
original source, which increases intelligibility and is more pleasant
to listen to.
Alternatives to two
loudspeakers
The simplest alternative to two loudspeakers is to dispense with
them entirely and to wear stereo headphones instead. People with
impaired hearing who resort in desperation to headphones (and to
restore peace with those around them) get an unexpected benefit as
the stereo spreads to a full 180 , and sound effects are disentangled
from speech. Headphones are also convenient where people of
different tastes are crowded together, for example on aircraft.
[3.5] Quadraphonic loudspeakers in
the home
A commercially short-lived system
using four loudspeakers: L, Left.
R, Right. LS, Left surround. RS, Right
surround.
With headphones the (broad) spread of the image is the same for
everyone, but several characteristics of the two-speaker system are
lost. One is that sound from both loudspeakers should reach both
ears, and do so from somewhere to `the front'; another is that the
sound stage is no longer static, so that if the listener turns it lurches
sideways. Headphones offer a different aural (and social) experience.
28
Stereo
The sound balancer should not use headphones to monitor stereo that
will be heard by loudspeaker. For the location recordist this may not
be avoidable, but the sound will usually be checked later on
loudspeakers.
A variation on two-speaker stereo is the 2.1 or 3.1 sound (also sold
to upgrade sound on home computers). The first figure designates
the number of frontal loudspeakers: in 3.1 a third is in the centre.
From conventional two-channel stereo, its output has to be derived
from the left and right components; if taken from 5.1 the track is
already available. The centre track helps to anchor sound to a
screen image.
5.1 surround sound is a full `jump' up from regular stereo. In
addition to left, centre and right at the front (LCR) it has extra
channels, left and right surround (LS and RS), towards the rear.
For the home this is supplied on CD or DVD-Video (both with a data
compression system that slightly reduces sound quality), or DVDAudio or SACD, Super Audio CD (both with full quality on a
suitable player). 5.1 requires a special sound mix (see Chapter 6),
comparable to the six-channel stereo originally developed for 70 mm
film and just a little more complex than the techniques used for
quadraphonics (`quad'), the old four-speaker system that failed
commercially. Surround then re-emerged in cinema with the centre
track to lock sound to picture. To allow for theatres that are very
wide at the back a further feed (or two) is added by matrixing extra
signals to those for the two rear-side speakers to fill in the rear from
an extra speaker or two. This is 6.1 or 7.1 surround.
[3.6] Loudspeaker array for 5.1
surround
L and R are equivalent to AB
two-channel stereo from the front,
and LS and RS (S for surround) are
placed at the rear sides of the listening
room: these four loudspeakers are
similar in function to those of `quad'.
C is added in the centre to coincide
with the screen for film, television or
video, or (if there is no picture) to give
a firmer location for sources in the
middle of the sound stage. Taking C
to be at 0 from the middle of the
room, recommended angles are 30
for L and R, and 110 (or perhaps
a little more) for LS and RS. The
position of the deep bass subwoofer
SW is not critical, as our perception
of direction is poor at very low
frequencies, and in any case it will
resonate throughout the enclosed
space. In theatres it will usually be
at the front. 2.1 has just L, R and SW.
3.1 has L, C, R and SW. 7.1 has
speakers on each side, like 5.1, and
two more at the back.
The `0.1' (for `one-tenth of the full frequency range') simply provides
deep (and potentially loud) bass ± typically 120 Hz and below ± from a
single, more powerful `subwoofer' loudspeaker. This can therefore be
of a separate design from the other five, which in turn benefit by their
isolation from its more extreme excursions and excessive harmonics.
The ears make little use of frequencies below 120 Hz to localize sound
sources, so a single speaker is sufficient for the sub-bass from any
direction. It will usually be off-centre (because the centre line is at a
node of many low frequencies), and often at the front, but otherwise
position does not matter much, provided nobody gets too close.
For cinema, including home cinema, the main uses of surround
sound are obvious: the subwoofer carries low-frequency effects
(LFE). Where 5.1 is used for music or for any other sound-only
application, the subwoofer receives extreme bass at frequencies that
are too low for the other five to handle. For music there are two main
ways in which the `surround' can be created ± just as for the
quadraphony (`quad') that preceded it, and which has left a substantial library of recordings.
The more conventional use for the five main channels is to place a
standard direct-sound stage between the three at the front, and
spread indirect sound (reflected sound, reverberation) to all five.
This offers the listener a sense of space that cannot be achieved by
the acoustics of a small room. In fact, most listeners probably prefer
that their room has this clear orientation, with a `front' and a `back'.
Stereo
29
The more adventurous alternative is to surround listeners with direct
sources, so they are placed in the middle of the action. To bring
individual sources forward, these can also be `placed' on the axis
between pairs of diagonally opposed loudspeakers. The effect can be
dramatic and exciting. Popular music can certainly make use of this.
Little existing classical music requires it, although exceptions include the Berlioz Requiem (with its quartet of brass bands), off-stage
effects in opera, and early antiphonal church music by composers
such as Gabrieli and Monteverdi.
Finally, for a fully 3-D sound field a system called `Ambisonics' has
been developed. This requires specialized microphone placement:
one design provides this within a single housing (the Soundfield
microphone; see Chapter 5). To handle the signal, it also requires
down-line techniques that are beyond the mainstream described in
this book.
Hearing and stereo
How, in real life, do our ears and brain search out the position of a
sound and tell us where to look for its source? First, the Haas effect
allows us to distinguish between direct and reflected sound (except
in the rare cases where the reflections are focused and their increased
loudness more than compensates for the delay in their arrival).
A second clue is offered by differences in the signal as it arrives at
our two ears: a sound that is not directly in line with the head arrives
at one ear before the other. At some frequencies the sounds are
therefore partly out of phase, and this may give some indication of
angle. Such an effect is strongest at about 500 Hz.
The physiology of the outer ear (the pinna) also helps to distinguish
front and rear, partly because it screens sound from behind. In
addition two ridges on the pinna create reflections with tiny delays
that can be processed by the brain. One of the ridges helps to some
extent to distinguish sound from above (providing some justification
for Ambisonics). Note here that one ear on its own can pick up a
significant amount of directional information: stereo is not restricted
to people with two good ears.
[3.7] Outer ear (the pinna)
This collects high-frequency sound
from the front and obstructs that
from the rear. In addition,
asymmetrical ridges on the outer ear
create faint reflections with delays of
80 ms or 100±330 ms in high-frequency
sounds from different directions.
Even with such tiny clues, a single ear
can often judge the direction of a
source, not just in a horizontal plane
but also in three dimensions.
But (using both ears again) the head itself begins to produce screening effects for frequencies at 700 Hz and above, resulting in differences in the amplitude of signals reaching the ears from one side or
the other. Further aids to direction finding are provided by movements of the head, together with interpretation by the brain in the
light of experience. Even with all these clues, with our eyes closed
we might still be uncertain: we can distinguish, separate out and
concentrate our attention on the contents of particular elements of a
complex sound field far better than we can pinpoint their components precisely in space. The clues tell us where to look: we must
then use our eyes to complete and confirm the localization ± if, that
is, there is any picture to see. If there is none, the fact that in real life
we rely rather heavily on the visual cues is perversely rather helpful
30
Stereo
to us: it permits us to continue to enjoy a spatially distorted image
even if we are a little off the stereo centre line.
Two-speaker stereo measures up to this specification quite well. To
understand how it does this, assume first that the source is in the
centre of the sound stage and that the listener is on the centre line.
The A signal reaches the left ear slightly before the B signal arrives,
and for middle and low frequencies the two combine to produce a
composite signal that is intermediate in phase between them. For the
right ear the B signal arrives first, but the combination is the same.
The brain, comparing the two combined signals, finds them to
be the same and places the source in the centre. If the amplitude of
the sound from one speaker is now increased and the other reduced,
the signals combine at the ears as before, but the resultant signals
differ from each other in phase ± in effect, it appears as though the
same signal is arriving at the two ears at slightly different times. This
was one of the requirements for directional information.
If the listener moves off the centre line, or if the volume controls of
the two loudspeakers are not set exactly the same, the sound stage is
distorted, but there is still directional information that matches the
original to some reasonably acceptable degree.
[3.8] Two-loudspeaker stereo: how
we perceive a central source
The signals from the A and B
loudspeakers are equal: the B signal
takes longer to reach the left ear, and
the A signal takes longer to reach the
right. But the combined signal C is
exactly in phase: the image is in the
centre.
[3.9] Stereophonic microphones
One way of picking up stereophonic
sound is the coincident pair: two
directional microphones very close
together (1) or in a common housing.
The two components will not
necessarily be at 90 to each other.
2, Main or `middle' and side
microphones used as a coincident
pair. 3, Symbol used in this book for a
coincident pair, which may be either
of the above.
Microphones for stereo
To create a stereo signal, the sound field has to be sampled by
microphones in such a way as to produce two sets of directional
information. There are several ways to do this. Given that the
loudspeakers used to reproduce stereo are separated, it might seem
logical to sample the field at points that are similarly spaced. However, the distance between the loudspeakers already produces phase
cancellation, and separating the microphones can only add to this.
Far greater control is achieved by placing them as close together as
possible, but pointing in different directions. They are then virtually equidistant from any sound source, and although (by their
orientation) they may pick up different volumes, the two signals
will be in phase for almost all the audio-frequency range. Only at
extreme high frequencies will a slight separation produce any
phase differences, and because this region is characterized by
erratic, densely packed overtones, even drastic phase changes will
be relatively unimportant.
This first arrangement is called the coincident pair, or coincident
microphone technique. There is a range of variations to it, but
typically the two microphones feeding the A and B information
are mounted one above the other or closely side by side and are
directional, so that the A microphone picks up progressively more of
the sound on the left and the B microphone more on the right. They
pick up equal sound from the centre, on a line halfway between their
directional axes. The frequency characteristics of the two microphones must be the same.
Stereo
31
An alternative kind of coincident pair uses radically different microphones, one behind the other. The front microphone covers most of
the sound field, including all required direct sound. It produces the
middle, main or `M' signal ± which, on its own, is also mono. The
second microphone picks up sound from the sides, producing the `S'
signal. This contains information required to generate stereo, but is
not, by itself, a stereo signal. Sounds from the two sides are opposite
in phase, so that if the M and S signals are added together they
produce a signal that is biased to one side of the sound field ± the
left ± and if S is subtracted from M it is biased the other way.
Quite simply, M + S ˆ A and M S ˆ B. To convert back, add
again: A + B ˆ 2M and A B ˆ 2B.
Whether by analogue or digital techniques, the conversion can be
made easily; indeed, its very simplicity can lead to confusion. For
example, if the two tracks of a stereo recording are added to make
a single-audio-track copy, this will work well if the originals were
A and B ± but if they were M and S, half the sound will
disappear.
[3.10] Two-loudspeaker stereo: how
we perceive an offset source
As the source is offset from the
centre to the left, signal A is
increased. The effect is to make the
peak in the combined signal C at the
left ear arrive a little earlier, and that at
the right ear later. This creates a slight
delay (t) between the two signals C.
The resulting phase difference is
perceived as a displacement of the
image to the left of centre. In these
diagrams it is assumed that the
listener is on the centre line between
A and B, and facing forward. Turning
the head will cause the phase
difference at the ears to change, but
in such a way as to keep the image
roughly at the same point in space. If
the listener moves away from the
centre line the image will be
distorted, but the relative positions of
image elements will remain
undisturbed.
There can be advantages to working with M and S signals, rather
than A and B. In particular, analogue A and B signals taking similar
but not identical paths may be changed differentially in volume or
phase; if this happens to the M and S signals, the perceived effect is
much less. Note also that because A + B will normally be greater
than A B, M should be greater than S, offering a useful visual
check (by meter) that the original signal is not grossly out of phase.
For video or film sound from a source within the visual frame,
M starts out at the centre of the sound stage, making it easy to shift
by any small amount.
Coincident pairs are only one way to pick up stereo. They are often
augmented by or may be replaced by spaced mono microphones.
Such spaced microphone techniques are widely used, and can also
give excellent results. Over the years, balancers who have been
prepared to experiment with competitive techniques have, at best,
improved the range of balances heard, but have also tended to
produce a cycle of fashions in the kinds of balance that are currently
in favour.
The only test that can be applied is practical, if subjective: does the
balance sound right? Experiments with a well-spread subject such as
an orchestra soon demonstrate that a single pair of widely spaced
microphones does not sound right: there is plainly a `hole in the
middle'. Sources right at the front in the centre sound farther away
than sources at the sides that are closer to the microphones. To
combat this, it is usual to add a third forward microphone in the
centre, the output of which can be split between the two sides. It also
helps to pin down the centre in surround sound, where there is a
central speaker.
One recording company's variation on this, the Decca `tree', has
employed three microphones in the middle: a splayed pair 2 m
(80 in) apart, with another 1.5 m (60 in) in front of them, pointing
forward.
32
Stereo
[3.11] Near-coincident pairs
Above: The ORTF array has
microphones separated by 17 cm
(6.7 in), about the width of a human
head. Below: In the NOS array
separation is 30 cm (12 in), about half
the circumference of a human head.
An alternative way of avoiding any hole in the middle of a spaced
pair is simply to move the microphones closer together. It is found
that if the spacing is less than 1 m (40 in), the problem is reduced,
and at two-thirds of that separation is generally cured. Some broadcasters have developed and used near-coincident layouts. These
include the Dutch (NOS) array that has cardioids angled outward
at 45 and spaced at 30 cm (12 in), and the French (ORTF) array of
cardioids angled outward at 55 and spaced at 17 cm (6.7 in). Also
on offer are other near-coincident arrays, for which separation, splay
angle and microphone are all precisely specified. These layouts have
been exhaustively analysed and justified, and there is no question
that they produce stereophonic effects that many find satisfying.
Anyone who works with a user will doubtless be directed to the
appropriate literature.
Yet another technique sometimes encountered is based on the idea of
a dummy head with pressure microphones in the position of the ears,
and the `sphere microphone' is a variation on this. In practice, the
`head' is often replaced by a disc that is designed to increase the path
difference for sound coming from the sides, and in this arrangement
is called a `baffled omni pair'. There certainly is a spatial effect,
because the A and B signals are slightly different, especially in the
high frequencies at which the obstacle effect of a real head helps to
define position. But the test is whether it works in practice and some
would argue that (except when heard on headphones, as for airline
in-flight entertainment) other techniques work better or are more
flexible.
It may also be argued that either coincident stereo pairs or spaced
mono microphones set closer in (or combinations of these) are
all that most readers will need. These are the main techniques
described here.
For radio or television broadcasters whose audiences have a mixture
of stereo or mono receivers, fully coincident stereo microphones
have a further virtue: the stereo signal from them translates easily to
mono. Spaced microphones are more problematical: as A and B
signals of similar strength from any source are combined in a mono
central channel, some frequencies (at regular intervals on a linear
scale) are enhanced; those in between are reduced: this is comb filter
distortion. In a complex sound, cancellations like these are so
common that they can be difficult to spot, but it seems unwise to
add to them.
[3.12] Baffled omni pair
Above: Sphere microphone, a
simplified `dummy head'. This has
two elements mounted in a spherical
baffle. Below: A baffle replaces the
sphere. In both arrangements the
obstacle increases time delays for
sound approaching from the side.
Although not widely used, it does
produce a stereo image.
Note also that microphones spaced in this way often transgress the
so-called `3:1 rule', which proposes that they should be placed
much further from each other than from the sources they cover.
(The 3:1 ratio arises from calculations for omnidirectional microphones: it ensures that when the sources covered are equally loud, a
signal that is picked up by the nearest neighbouring microphone
will be at least 10 dB lower.) Balancers seem to observe this rule
with more care when setting up spotting microphones for music (see
Chapter 8).
Stereo
33
Stereo compatibility
If a stereo signal produces acceptable mono sound when reproduced
on mono equipment, it is said to be compatible. As we have seen, the
M signal (or A + B) may be compatible, but this is not automatic: it
would be easy to place an M and S microphone in such a way that it
is not. To achieve compatibility, the M component should include all
the required direct sound sources in acceptable proportions, usually
together with enough indirect sound (reverberation) to produce a
pleasing effect, but not so much as to obscure it.
Domestic listening conditions vary greatly: in particular, the stereo
component may be debased or absent. To allow for this, compatibility is desirable. Only where the listening conditions can be
controlled is compatibility dispensable, but this means also that the
material cannot be transferred for other uses without further, special
processing.
[3.13] Spaced pair
Above: Two microphones. For
sources along the centre line their
apparent distance from the front of
the audio stage is greater than for
sources in line with the microphones;
and as the source moves forward,
this `hole-in-the-middle' effect is even
more pronounced. Below: This is
filled in by adding a third microphone
in the centre.
Historically, many analogue stereo systems have been arranged for
automatic compatibility: for example, on stereo records. On a vinyl
disc, A + B corresponds to lateral displacement of the stylus, so a
mono signal is easy to pick up. A B is recorded as vertical hilland-dale signals: a mono stylus rides up and down on it but cannot
convert that into an electrical signal. Similarly, on analogue magnetic tape the stereo A and B signals are side by side on separate
tracks. A full-track head reproduces both together, or alternatively
the combined signal can be obtained electrically by adding the two
outputs.
Similar advantages are obtained from using A + B and A B signals
in radio. When stereo was introduced, the A + B signal occupied a
place in a waveband previously used for the mono signal; the A B
signal was modulated on to a subcarrier above the main A + B band.
Today, digital systems have replaced many of these intervening
stages. Provided that the A and B or M and S signals are interlocked
so they cannot be displaced, they can be converted back into an
analogue system that preserves compatibility. For 5.1 surround
sound to be compatible with stereo and mono it requires special
features such as those devised by Dolby.
`Spatialization', using out-of-phase effects to increase the apparent
width of stereo (to more than the speaker separation), is incompatible with mono because cancellation effects appear. It is not recommended here as a studio technique, but for domestic use adds
spaciousness, at the cost of some confusion of the intended stereo
image.
Chapter 4
Studios and acoustics
At the heart of most studios there is a dedicated acoustic environment
within which sounds are created, then picked up by microphones.
This space is designed or arranged to serve a particular purpose or
function, with the layout, acoustics, treatments and furnishings that
are required to achieve it. The studio may be used for radio, or for
recordings (often of music), or for television. A film stage (not `film
studio', which is the whole organization) is dedicated to successive
individual shots and has less electronic equipment, so although
similar is somewhat simpler than the television studio. (Film and
television re-recording theatres and MIDI-equipped `virtual' studios,
where acoustics play a lesser role, are described in later chapters.)
The music recording industry alone is so vast that it has generated
its own great range of studio variants. Many of the more specialized
books on professional audio are designed primarily to serve the
practitioners of this one group; here, however, a broader range will
be covered. But acoustics is such a broad field that its principles
apply with only minor changes of application to almost any kind
of studio.
The range and function of
studios
Broadcasting studios vary according to the economic needs and
the size of the country, region or community they serve, and with
the way they are funded. The cost of television limits the number
of well-equipped studios that a given population can support. A
regional population of five to eight million is generally sufficient
to sustain several centres, each with about three studios for major
productions (including drama), for general purposes (including
Studios and acoustics
35
entertainment, with audience), and for speech only (presentation and
news), plus an outside (remote) broadcast unit.
A smaller population makes it difficult to sustain even this number:
there may be only two television studios or even one. There would also
be less use for an outside broadcast unit, which has in any case been
partly displaced by the use of individual video cameras (camcorders).
At the lower end of the scale, towns serving areas of a million or so
may have single-studio stations (or more, in prosperous regions).
But these studios will be small, big enough for only two or three
setting areas of moderate size.
A national population the size of the UK or the USA is sufficient to
sustain four or more national broadcast networks, augmented by a
large, and seemingly ever-increasing, range of local and cable or
satellite channels.
In radio (existing, in most countries, in the shadow of television),
an even more complicated situation exists. In the USA there is a radio
station to every 25 000±30 000 population. Of the vast total number,
half subsist on an unrelieved output of `middle-of-the-road', `conservative' or familiar `wall-to-wall' music in some narrowly defined
style. Then there are the top 40, the country-and-western and the
classical specialists, plus stations that offer just news, conversation
(e.g. telephone chat) or religion. Educational stations have a broader
range, broadcasting both speech and music. In addition, a few
`variety' stations still exist, segmented in the old style and containing
an integrated range of drama, documentary, quiz, comedy, news and
so on. There are also ethnic stations that serve the broader interests
of a defined community. Most of these survive with low operational
costs by using records or syndicated recordings for much of their
output, and their studio needs are necessarily simple: perhaps no more
than a room converted by shuttering the windows against the lowfrequency components of aircraft noise that heavy drapes will not
impede. Acoustic treatment can be added on the inside of the shutters.
For some stations, a second, general-purpose studio may be partly
justified by its additional use for board or staff meetings.
But looking again at the region with five to eight million people, we
now have a population that can sustain a full range of studios for all
of the specialized functions that the current purposes of radio permit.
These are likely to be fewer in the USA and other countries that have
followed its example in dedicating commercial stations to a narrow
range of output, than in public service broadcasting, where a regional
centre may have the following studios:
.
.
.
.
.
speech only (several);
pop music;
light and orchestral music;
light entertainment (with audience);
general purpose, including drama and education.
Smaller regions (100 000 to one million population) are capable
of sustaining vigorous, but more limited, radio stations. National
populations are capable of sustaining networks in about the same
36
Studios and acoustics
number as for television (though these are much more fragmented
in the USA than in the UK, where they all reach almost the entire
population). The BBC has about 60 radio studios in London alone,
though many are for overseas broadcasting. Some are highly
specialized: for example, a few that are acoustically furnished for
dramatic productions in stereo, a form that still exists in Britain.
Noise and vibration
To help in designing or choosing buildings for radio, television
(or any other sound studio), the first requirement is acoustic isolation.
Also, if noise cannot get in, neither will it get out and disturb the
neighbours. Here are some general guidelines:
.
.
.
.
.
.
.
.
.
.
Do not build near an airport, railway or over underground lines.
Prefer massive `old-fashioned' styles of construction to steelframed or modern `component' architecture. Thick brick or concrete
walls are good; wood or plasterboard much less so (but if they are
used, adhesives are better than nails).
In noisy town centres, offices that are built on the outside of the
main studio structure can be used as acoustic screening. These
outer structures can be built in any convenient form, but the studio
must be adequately insulated against noise from (or to) them.
The best place to put a studio is on solid ground.
If a radio or recording studio cannot be put on solid ground, the
whole massive structure, of concrete floor, walls and roof, can be
floated on rubber or other similar materials. The resonance of
the whole floating system must be very low: about 10 Hz is
reasonable. One inch of glass fibre or expanded polystyrene gives
resonances at about 100 Hz, so care is needed. An alternative is to
suspend the entire structure. If floors or false ceilings are floated
separately, they should not be fixed to side walls.
Airborne noise from ventilation is a major problem. Low-pressure ducts are broad enough to cause resonance problems if not
designed with care: they also require acoustic attenuation within
them to reduce the passage of sound from one place to another.
Grilles can also cause air turbulence and therefore noise.
High-speed ducts have been suggested as an answer to some of the
problems of airborne noise, but would also require careful design,
particularly at the points where the air enters and leaves the studio.
Double doors with pressure seals are needed.
Holes for wiring should be designed into the structure and be as
small as possible. They should not be drilled arbitrarily at the
whim of wiring engineers. Wiring between studio and listening
room should pass round the end of the wall between them,
perhaps via a corridor, lobby or machine room.
Windows between studio and control should be triple or possibly
double glazed, with several glass thicknesses, say 6 and 9 mm
(1/4 and 3/8 in), and the air space between pairs at least 100 mm
(4 in). Contrary to what might be expected, it does not seem to
Studios and acoustics
.
[4.1] Permissible background noise
in studios
1, Television studios (except drama
and presentation). 2, Television
drama and presentation studios and
all radio studios except those for
drama. 3, Radio drama studios.
.
37
matter much whether they are parallel, although in practice one is
often angled by 5 or more. Triple glazing may not be necessary if
the loudest sounds are likely to be much the same on both sides of
the glass. Prefer elastic (e.g. rubber) seals around the edges.
Windows to areas outside the studio area or to the control area if
the loudspeaker is to be switched to replay during recordings
should be triple glazed if replay or noise levels are likely to
be high. Measures to avoid condensation between the glass
panes will be necessary only in the case of windows to the outer
atmosphere, where there will be temperature differences.
Any noisy machinery should be in a structurally separate area,
and on anti-vibration mountings.
Massive structures are recommended, but to make them extremely
massive may be disproportionally expensive. Doubling the thickness
of a brick wall reduces transmission by only 5 dB.
For television studio roofs the lower limit for single-skin construction is about 25 cm (10 in) thick: this gives an average attenuation
of 60 dB over the range of 100±3200 Hz. For noisy jet aircraft
flying at 1000 ft, a minimum of 65 dB attenuation is desirable. Even
low-flying helicopters are less of a problem than this (provided they
do not land on the studio roof), but supersonic aircraft producing a
sonic boom with overpressures of 9.6 kg/m2 (2 lb/ft2) would require
70 dB attenuation. Since a double skin is in any case desirable for
more than 60 dB attenuation, it may be wisest to adopt this higher
standard ± of 70 rather than 65 dB attenuation. This results in
a structure that is little heavier than a 60 dB single skin.
Subjective tests have been carried out using various types of aircraft
noise, attenuated as if by different kinds of roof, then added to
television recordings at quiet, tense points in the action (in the
limiting case the action may be in period drama, where aircraft noise
would be anachronistic). Obviously, permissible noise levels of any
kind depend on the type of production. In television drama, background noise from all sources, including movement in the studio,
should ideally not exceed 30 dB (relative to 2 10 5 Pa), at 500 Hz.
For light entertainment the corresponding figure is 35 dB. However,
in practice, figures up to 10 dB higher than these are often tolerated.
Reverberation
The basic studio structure is a bare empty box of concrete (or some
similar hard and massive material). It will reflect most of the sound
that strikes it, and its capacity to reflect will be almost independent
of frequency, though very broad expanses of flat thin concrete may
resonate at low frequencies and so absorb some of the low bass.
When typical furnishings are placed in a room they include carpets,
curtains and soft chairs. These (together with people) act as sound
absorbers. Other furnishings include tables, hard chairs, and other
wooden and metal objects. These reflect much of the sound striking
38
Studios and acoustics
them, but break up the wave fronts. They act as sound diffusers. Some
things do both: a bookcase diffuses sound and the books absorb it.
[4.2] Early reflections
After direct sound (1) is heard, there
is a short delay, then a number of
early reflections help to define the
character of the acoustic environment
(2). With more reflections, this
develops into densely structured
reverberation (3). If the gap before
the first reflection is more than 55 ms,
it will be heard separately as an echo,
but if subsequent reflected sound has
shorter delays the effect should be
pleasant.
The absorption or diffusion qualities of objects vary with frequency. In
particular, the dimensions of an object condition how it behaves. A
small ornament diffuses only the highest frequencies; sound waves that
are long in comparison to its size simply pass round it. The thickness of
a soft absorber affects its ability to absorb long wavelengths, and so
does its position. At the hard reflecting surface of a rigid wall there is
no significant air movement, so 2.5 cm (1 in) of sound-absorbing
material reaches out into regions of substantial air movement (and
damps them down) only for the highest frequencies; to have any effect
on lower frequencies, the absorber must be well away from the wall.
Sounds in an enclosed space are reflected many times, and some
(great or small) part of the sound is absorbed at each reflection.
The rate of decay of reverberation defines a characteristic for each
studio: its reverberation time. This is the time it takes for a sound
to die away to a millionth part of its original intensity, i.e.
through 60 dB. Reverberation varies with frequency, and a studio's
performance may be shown on a graph for all audio frequencies.
Alternatively it may be given for, say, the biggest peak between 500
and 2000 Hz, or at a particular frequency within that range.
Reverberation time depends in part on the distance that sound must
travel between reflections, so large rooms generally have longer
reverberation times than small ones. This is not only expected but
also, fortunately, preferred by listeners.
In between the first (direct) sound and the densely bunched muchreflected sound that we think of as reverberation, there is a period when
the earliest reflections arrive from the main walls and other large surfaces
on the studio. The time to the first reflection, and the way other early
reflections are spread in time, also help to define studio quality. (They
are also important when creating or choosing artificial reverberation.)
After a few reflections, reverberation is spread fairly evenly and therefore is at much the same volume throughout an enclosed space, in
contrast to direct sound, which dies away as the inverse square of the
distance from the source. In large reverberant halls, at some critical
distance the direct sound is overwhelmed by reverberation. Beyond the
critical distance, the intelligibility of speech is reduced, then lost, and
melody is submerged in mush. Theatres need to be designed so that no
one has to sit near or beyond this critical distance, and microphones set
to balance direct and reverberant sound will always be well within it.
Coloration
[4.3] Critical distance
S, Sound source. D, Direct sound
dominant. C, Critical distance.
R, Reverberation dominant.
In a large room an echo may be detected. If there is little reverberation
in the time between an original sound and a repetition of it, and if this
time gap exceeds about an eighteenth of a second, which is equivalent
to a sound path of some 18 m (60 ft), it will be heard as an echo.
Studios and acoustics
39
In a small room, coloration may be heard. This is the selective
emphasis of certain frequencies or bands of frequencies in the
reverberation. Short-path reflections (including those from the ceiling) with less than 20 ms delay interfere with direct sound to cause
cancellations at regular intervals throughout the audio-frequency
range, resulting in a harsh comb filter effect.
In addition, there may be ringing between hard parallel wall surfaces
which allow many reflections back and forth along the same path: at
each reflection absorption always occurs at the same frequencies,
leaving others still clearly audible long after the rest have decayed.
Eigentones, the natural frequencies of air resonance corresponding
to the studio dimensions, are always present, and in small rooms are
spaced widely enough for individual eigentones to be audible as
ringing. If the main dimensions are direct multiples of, or in simple
ratios to, each other, these may be further reinforced.
[4.4] How sounds die away
1, In a good music studio the sound
dies away fairly evenly. 2, In a poor
music studio (or with bad microphone
placing) the reverberation may decay
quickly at first, and then more slowly.
The rate of absorption in different parts of a hall may not all be the
same, giving rise to anomalous decay characteristics. For example,
sound may die away quickly in the partially enclosed area under a
balcony, but reverberation may continue to be fed to it from the main
hall. More unpleasant (though less likely) might be the reverse of
this, where reverberation `trapped' in a smaller volume is fed back to
a less reverberant larger area. Resonances may interchange their
sound energies: this happens in the sounding board of a piano and
is one of the characteristics that we hear as `piano quality'. In a room
it may mean that a frequency at which the sound has apparently
disappeared may recur before decaying finally.
Not all of these qualities are bad if present in moderation. They may
give a room or hall a characteristic but not particularly unpleasant
quality: something, at any rate, that can be lived with. In excess they
are vices that may make a room virtually unusable, or usable only
with careful microphone placement.
The answer to many of these problems ± or at least some improvement ± lies, in theory, in better diffusion. The more the wave fronts
are broken up, the more the decay of sound becomes smooth, both in
time and in frequency spectrum.
In practice, however, if the studio geometry has no obvious faults
and if acoustic treatment has been carried out and if panels of the
various types that may be necessary for adequate sound absorption
have been distributed about the various walls and on the ceiling,
little more generally needs to be done about diffusion.
Studios for speech
Studios used only for speech are often about the same size as a living
room in an ordinary but fairly spacious house. Where lack of space
leads to smaller rooms being used, the result may be unsatisfactory,
with awkward, unnatural-sounding resonances that even heavy acous-
40
Studios and acoustics
tic treatment will not kill. And, of course, acoustic treatment that is
effective at low frequencies reduces the working space still further.
Coloration, which presents the most awkward problem in studio
design, is at its worst in small studios because the main resonances
of length and width are clearly audible on a voice containing any
of the frequencies that will trigger them. An irregular shape and
random distribution of acoustic treatment on the walls both help to
cut down coloration. But in small rectangular studios treatment may
be ineffective, and this means working closer to the microphone.
With directional microphones, bass correction must be increased and
the working distance restricted more than usual.
Coloration is not necessarily bad (a natural-sounding `indoor' voice
always has some) but it has to be held to acceptable limits ± which
are lower for monophonic recordings. Even on a single voice there is
great difference between mono and stereo, because with mono all of
the reverberation and studio coloration is collected together and
reproduced from the same apparent source as the speech. Note that
even when the final mix is to be in stereo, pre-recorded voices are
often taken in mono, so the acoustics of (for example) a commentary
booth should allow for this.
One solution might appear to lie in creating an entirely `dead' studio,
leaving only the acoustics of the listener's room. But listening tests
indicate that most people prefer a moderately `live' acoustic,
whether real or simulated. For normal living-room dimensions,
0.35±0.4 s seems about right for speech when using directional
microphones; a smaller room would need a lower time, perhaps 0.25 s.
Given that reverberation time is defined as the time it takes for a
sound to die away through 60 dB, you can use a handclap to give a
rough guide to both duration and quality. In a room that is to be used
as a speech studio the sound should die away quickly, but not so
quickly that the clap sounds muffled or dead. And there must
certainly be no `ring' fluttering along behind it.
Listening rooms and sound-control cubicles should have similar
acoustics, with a maximum of 0.4 s at frequencies up to 250 Hz,
falling to 0.3 s at 8000 Hz. Treatment might be distributed to give
good diffusion at the sides and greater absorption at the rear. Larger
theatres (and particularly cinemas) need to be dead enough to ensure
that sound at the back is not dominated by reverberation: in such
conditions, film dialogue may be clearer than when heard in a more
lively home acoustic.
General-purpose sound studios
[4.5] Radio studios
Typical reverberation times for:
A, Programmes with an audience.
B, Talks and discussions. C, Classical
music. D, Drama, popular music.
In some countries there is still a need for studios where a variety
of acoustics is used for dramatic purposes. In others, such facilities
can hardly be justified except, perhaps, to add interest to commercials. They may also be used for special projects to be issued as
Studios and acoustics
41
recordings. This studio type is therefore still important, but less than
when radio was the principal broadcasting medium.
[4.6] Television studios
Typical reverberation times for
general-purpose studios.
In a dedicated drama studio there are two main working areas, usually
separated by no more than a set of thick curtains. One part, the `dead'
end, has a speech acoustic that is rather more dead (by about 0.1 s)
than would be normal for its size. The other end is `live', so that when
used for speech the acoustic is obviously different from the dead end.
If the curtains are open, a speech microphone in the dead end can
sometimes pick up a `live' resonance ± especially if a directional
microphone is open towards the live end. This parasitic reverberation
generally sounds unpleasant unless there is a good reason for it, so the
curtains are usually drawn far enough to screen off the live end. But
useful effects can sometimes be found near the middle of the studio.
The live end can also be used for small music groups working with
speech taken from the dead part of the studio.
[4.7] Range of reverberation times
A, Median for any particular type.
Below 200 Hz the reverberation in
talks studios and their control rooms
may rise; above 3150 Hz, for most
studios it should fall by 10±15%. For
a large music studio the tolerance is
about 10% on either side of the
average; for other studios and control
rooms it may be a little wider.
There may also be a very dead area, although if this is small it may
sound like the inside of a padded coffin rather than the open air it is
often intended to represent. For good results the type of treatment
used in anechoic chambers for sound testing would be best for the
walls and ceiling: these have wedges of foam extending out about
1 m (3 ft or more) from the walls. On the floor a good carpet
(with underlay) is sufficient. Note that some performers find these
conditions unpleasantly claustrophobic.
A studio for stereo speech (or any space within a studio that is used
for this) needs to be more dead than other general-purpose studios.
The reason for this is that where coincident microphones are used,
voices must be placed physically farther back than for mono; otherwise, any movement will be exaggerated.
Music studios
For many audio professionals, the production of music, and very
little else, is what studios are for. But `music studio' means some
very different things to different people: it can be live, making full
use of rich acoustics; it may be dead, with the acoustics synthesized;
it can even, in fact, not be there at all, present only as facilities
in a `virtual studio' (see Chapter 16). Many music studios have
combinations of these characteristics. To some extent the choice
depends on the type of music, but fashion also seems to play a part.
At one time, almost all popular music recordings were made in very
dead acoustics; since then, there has been a shift back to include some
live acoustic components, and studio design now often reflects this.
A primary characteristic of any music studio with live acoustics is
its reverberation time. Preference tests show that, for any given size
of studio, music requires longer reverberation than speech. Such
preferences are not absolute, but are valid for most conventional
music and instruments. As we know them in the West today, these
42
Studios and acoustics
have developed in a very specialized way: in particular, our orchestral and chamber music has developed from that played in rooms
within large houses in seventeenth- and eighteenth-century Europe.
The original acoustics defined the music ± and that music now, in
turn, determines the acoustics needed for it.
In contrast, at one extreme, much Eastern music is designed for
performance in the open air, as also is much military or brass-band
music. Also, some modern music is written for drier acoustics than
those that sound best with earlier classical works. Nevertheless, the
acoustics we have been accustomed to have in their turn led to the
design of instruments for which music of great beauty and power has
been composed; so the studios we look for have certain specific
qualities which can be judged by these subjective criteria.
Listening tests are not always the best indication of true value: they
measure what we like best at the time of the test, and not what we
could learn to like better if we chose to change our habits. But it does
seem clear that the ideal (i.e. preferred) reverberation time varies
with the size of studio: small studio ± short reverberation; large
studio ± long reverberation. For the very smallest studios (e.g. a room
set aside as a music room in a private house) the ideal is between
0.75 and 1 second. This is what might be expected if the room has
no carpet and few heavy furnishings. Some authorities suggest that
there should be more reverberation in the bass.
An interesting sidelight on this is that, on old mono recordings, it
sounded better if the reverberation was about a tenth of a second less
than that for live music. When reverberation came from the same
apparent source as the direct sound, less seemed to be required.
Apart from reverberation, the principal quality of a good music
studio is good diffusion, so that the sound waves are well broken
up and spread. There should be no echoes due to domes, barrel
vaults or other concave architectural features that can concentrate
sound.
Dead pop music studios
There was a phase in popular music production when studio acoustics were little used. The internal balance of the music was deliberately surrendered in exchange for electronic versatility, and here
nothing has changed. A celesta, piano or flute might be given a
melody that has to sound as loud as brass, or other effects may be
sought but can be achieved only if the sound of a louder instrument
does not spill to the microphones set at high amplification for quiet
instruments. Where separation is prized above natural acoustics,
microphones are placed as close to instruments as may be practicable, and their directional properties are exploited to avoid picking
up direct sound from distant loud sources, and to discriminate
against reverberation from the open studio. In this musical style,
reflected sound is just a nuisance.
Studios and acoustics
43
A deliberately dead music studio was made a great deal more dead
than most speech studios. Such an acoustic environment requires
strong attenuation even at a single reflection from any wall, so that
much less of the sound of (for example) the brass will bounce off the
walls to reach the flute or celesta microphone from directions that
cannot be discriminated against.
An aspect of acoustics that is important to pop music, not for its
presence but again by its absence, is attenuation in air. At frequencies
below about 2000 Hz, absorption by the air is negligible. But for
8000 Hz, air at 50% relative humidity has an absorption of 0.028
per 30 cm (per foot). This means that for high frequencies there is a
loss of about 3.5 dB at a distance of 30 m (100 ft). For damper air
the loss is a little lower, while for very dry air it rises sharply and
begins to affect lower harmonics (although still mainly over 2 kHz)
appreciably. For classical music this can make a big difference
between performances using a distant balance in the same concert
hall. But it does not affect close-balanced pop music, except to make
instrumental quality different from that which may be heard acoustically: more of the higher harmonics of each instrument reach the
microphone.
Television and film studios
Studios for television and film (and particularly the larger ones) are
usually made very dead for their size. The primary reason is that if
they are not to appear in vision, microphones must be farther away
from sources than in other types of balance. But it also helps to
reduce inevitable noise from distant parts of the studio. Noise arises
in a variety of ways. Studio lamps produce heat, and the ventilation
for that cannot be completely quiet. Scenery may have to be erected
or struck during the action, or props set in. Performers and staff must
walk in and out of the studio, and from set to set. Cameras move
about close to the action, dragging cables. The floor of a television
studio must allow the smooth movement of heavy equipment, so is
acoustically reflective.
Sound that radiates upward may be reflected several times before it
can reach a microphone, but sound that travels horizontally might be
reflected only once. For this reason the most effective place for the
absorption of sound is in the lower parts of the walls; more treatment
is concentrated here than is placed higher up.
[4.8] Four television studios:
frequency response
Variation in reverberation time with
frequency. BBC Television Centre
Studios 2 and 5 are both 3320 m3
(116 000 ft3); studios 3 and 4 are
10 230 m3 (357 000 ft3). Studio 4
was designed primarily for
musical productions.
The set itself may reduce the overall reverberation a little: braced flats
absorb low frequencies, while soft furnishings and thick drapes reduce
the highs. In addition, the set may have unpredicted effects on the
local acoustics: hard surfaces may emphasize high frequencies. A cloth
cyclorama makes little difference, but if it is constructed from wood
or plaster it reflects sound strongly (and curved parts focus it).
Television studios have acoustics approaching those used in `dead'
music studios, so they are suitable for music techniques of this type.
44
Studios and acoustics
However, for the occasions when an orchestra has to be brought into
the studio, the liberal application of unheard artificial reverberation
is no help to musicians who find the surroundings acoustically
unpleasant. In particular, string players need reflected sound to hear
that they are producing good tone quality at adequate power. In its
absence they may bow harder, changing the internal balance of the
orchestra and making its tone harsh and strident. The timing of
ensemble playing also suffers. An orchestral shell (an acoustically
reflective surface behind and above) helps all the players, but
reinforces the brass players (who do not need it) more than anyone
else: a full shell turns a balance inside out and should not be used.
It is best to design the set with reflecting surfaces to benefit the
strings: this does nothing for the overall reverberation time, but
helps musicians to produce the right tone.
Film studios (`stages') have in the past tended to be even deader than
television studios. In addition to the other problems, film cameras
are, by the nature of their intermittent action, noisy and difficult to
silence effectively, while some HD video cameras (with nearby
`video villages') can be even worse.
Television sound control rooms and film dubbing theatres have the
same characteristics as their counterparts in radio: 0.4 s at 250 Hz
falling to 0.3 s at higher frequencies. The acoustics of the main
production and lighting control areas should be relatively dead, in
order that the director and his assistant and the technical director can
be clearly audible on open microphones in areas that also have
programme sound. In practice, however, absorption is limited by
the necessarily large areas of hardware and glass.
Other technical areas, such as those for telecine and recording
equipment, may need acoustic treatment: some machines can be
noisy. If these are grouped in partly open bays, acoustic partitioning
is desirable: aim for 20 dB separation between one bay and the next.
Acoustic treatment
There are several possible types of acoustic treatment:
[4.9] Sound absorbers
Padding that is close to a reflecting
surface is efficient only for short
wavelengths. Screens with a thin
layer of padding are poor absorbers
at any but the highest frequencies.
For greater efficiency in absorbing a
particular wavelength, the absorber
should be placed at a quarter
wavelength from reflecting surfaces.
In studios where space permits,
boxes of different depths may be
sunk into the wall, with quarterwavelength absorber inside them,
flush with the walls.
Soft absorbers. These are porous materials applied to walls: their
effectiveness depends on loss of sound energy as air vibrates in
the interstices of the foam, rock wool or whatever is used. The
method works well at high and middle frequencies, but for efficiency at low frequencies absorbers need to be about 4 ft (1.2 m)
deep. So it is reasonable to lay absorbers at thicknesses that are
efficient down to about 500 Hz, then become less so. Many types
are commercially available. Control of high frequencies (for which
absorption may be excessive) can be moderated by using a perforated hard surface (e.g. hardboard): with 0.5% perforation much of
the `top' is reflected; with 25% perforation much of it is absorbed.
By changing the proportions of these two surfaces, the acoustics may
be customized.
Studios and acoustics
45
[4.10] Absorbers
The membrane absorber (left) is
a box covered by a panel that
is fixed at the edges and
weighted with some material
such as roofing felt. It absorbs
a broad range of frequencies (1).
The Helmholtz or damped
cavity absorber (right) is in
effect a bottle of air resonating
at a particular frequency, with
damping material inside it.
It absorbs at the frequency
of resonance (2).
Helmholtz resonators. These are not widely used, but the principle is
available. Cavities open to the air at a narrow neck resonate at
particular frequencies. If the interior is damped, they absorb at that
frequency. Such resonators can be used for attacking dimensional
resonances in studios or halls. A potential disadvantage is that it
requires treatment to be fitted either at particular places that may
already be occupied or some of the only places that are not.
Membrane absorbers. These have sometimes been used to cope
with excessive low frequencies ± although if low frequencies are
a problem, the absorbers can be too. The idea is simple enough:
a layer of some material or a panel of hardboard is held in place
over an air space. The material moves with the sound wave like a
piston, and this movement is damped so that low-frequency sound
energy is lost.
A combination absorber might consist of units of a standard size:
say, 60 cm 60 cm 18 cm deep (2 ft 2 ft 7 in). These are easy
to construct and install; building contractors can be left to it with
little supervision by acoustical engineers. A level response for a
studio can be obtained by making (literally) superficial changes to
this single design.
[4.11] All-purpose absorber
A plywood box mounted on battens
fitted to the wall and covered with
rock wool 3 cm thick (about 11/4 in)
and perforated hardboard. The
response is tuned by varying a single
factor, the amount of perforation in
the hardboard surface layer.
Combinations of boxes with 0.5%
and 20% perforations can be used
as required.
The box is made of plywood, with a plywood back. Hardboard
dividers are used to partition the interior into four empty compartments 15 cm (6 in) deep. Over this airspace, and immediately behind
the face of the box, is laid 2.5 cm (1 in) of heavy-density rock wool.
The face itself should be made of perforated hardboard or similar. If
the open area perforation is very small there is a resonance peak of
absorption at about 90 Hz. If the perforation area is large (20% or
more), wideband absorption is achieved, but with a fall-off at 100 Hz
and below. By using similar boxes with two kinds of facing, it is
possible to achieve almost level reverberation curves.
Decorative diffusers can also be specially constructed, perhaps for the
sides of a control room. A tray with sides and ends is made by using
wood 15 or 20 cm deep by 25 mm thick (6±8 in 1 in). The length of
the tray could be rather more than half the height of the room. The
width of the tray is divided at regular intervals by more strips of the
same wood, to make a series of deep slots. The slots are given a range
of different depths by separating the dividers with narrower strips of
wood. In fact, the best way to make the tray might be to work from
46
Studios and acoustics
one side to the other, then fix the ends ± i.e. top and bottom ± and
finally panel the back. The whole tray can then be supported about
30 cm (1 ft) or so above floor level, and away from the wall (so that a
layer of damping absorber fits behind) on floor to ceiling battens set
back at the sides. These in turn can be used to mount surface layers of
absorbers below, above and to the sides of each diffuser.
Using sound absorbers
[4.12] Diffuser
1, Part of wooden diffuser (in plan
view). 2, Support strut running from
floor to ceiling. 3, Studio or control
room wall. 4, Space for acoustic
damping behind diffuser. 5, Front of
absorber: this fills the spaces
between neighbouring diffuser
panels, and above and below them.
If the acoustics of a sound studio are fully designed, a variety of
absorbers is likely to be used. To help evaluate them, their absorption
coefficients are measured. That most commonly used is averaged over
all angles of incidence, and over the range from 100 to 3150 Hz.
Where there is total reflection, the absorption coefficient is zero; for
total absorption it is unity. In practice the figure should therefore lie
between zero and one. For example, free-hanging heavy drapes might
be 0.8 for 500 Hz and above. But note that for relatively small areas
absorbers may have a coefficient that is greater than one ± an apparent
impossibility that can be produced by diffraction effects.
Larger objects have sometimes been described in terms of the
number of `open window units' (their equivalent in the number of
square feet of total absorption) that they contribute. On this scale,
individual, separated people check in at about four and a half units
(at lower-middle to high frequencies), and a padded sofa may have
10 times the effect of the person sitting on it. For seated audiences,
this would give too high a number, as sound does not approach
each individual body from all angles, so it is necessary and in any
case more convenient to calculate their absorption as though they
are part of a flat surface: for occupied seating this gives a coefficient of 0.85.
Tables of these values are available: in addition, they give values for
the studio structure as well and for the volume of air within it. The
calculations are complex, but here are some rule-of-thumb ideas that
might help when trying to correct the acoustics of a given room:
.
.
.
apply some of each type of absorber to surfaces affecting each of
the three dimensions;
try to ensure that no untreated wall surfaces remain to face each
other (note that a surface that is treated at one frequency may be
reflective and therefore effectively untreated at another);
put the high-frequency absorbers at head level in speech studios.
The cheapest and best treatment for the floors of speech studios is a
good carpet and underlay (but the roof will need compensation in the
bass). Wooden blocks are good on the floor of a (live) music studio.
Take equal care with furnishings. Plainly, it is unwise to get the
acoustics right for an empty studio and then to bring in a lot of items
(and particularly, any large soft absorbers) that change it. Furniture,
decoration and even people must be included in the calculations
Studios and acoustics
47
from the start. But apart from this, it is not worth worrying too much
about problems until you actually hear them. For example, if there is
enough treatment in a room to get the reverberation low and level,
and if the main rules for the layout of absorbers have been observed,
diffusion of the sound will probably be all right anyway. So you can
arrange the furniture for how it looks best, makes the occupants
comfortable, and gives the most convenient layout of working areas.
The use of screens
Very often the acoustics that are built into a studio are not exactly
those that are wanted for a particular use, so ways of varying studio
quality have been devised. Sometimes the acoustic treatment of a
studio is set in hinged or sliding panels, so an absorber can
be replaced by a reflecting surface. But a more flexible approach
relies on layouts of studio screens ± in America, called flats or gobos
(`go betweens').
If padded screens are to be easily movable (on small wheels), they
must be light in weight (which in practice means they cannot be very
thickly padded) and not too broad, often about 0.9 m (3 ft) wide. For
stability they stand on cross-members at each end or on a tripod-like
mount in the middle. Individual free-standing screens are effective
only for sound of short wavelengths: low-frequency sound flows
round and past them as though they were not there. But grouping
screens edge to edge improves on this. At best, thin screens
may reduce the sound flow between neighbouring areas by some
10±15 dB, while heavier screens with more absorber, say 10 cm
(4 in) thick, will do better.
A common type of screen consists of a panel of wood with a layer
of padding on one side. Placed with the dead side to the microphone,
it will damp down the highs. With the bright (reflective) side
forward, the highs are emphasized, while ambient sound in
the studio is somewhat reduced by the absorbent back. In music
studios with dead acoustics, large free-standing sheets of Perspex
or wood are used in addition to the thickly padded screens
that separate instruments and their associated microphones from
each other.
Whatever the type, the effect of all screens is the same in several
important respects:
.
.
With screens to get in the way, there are shorter path lengths
between reflections of the sound in the studio, so there are more
reflections and more losses in a given time. This leads, in general,
to a lower reverberation time for the studio as a whole.
Sound from distant sources is reduced overall, but sound that
does get round the screens is usually lacking in top. This also
applies to any appreciable reverberation from the open studio.
48
Studios and acoustics
.
.
Coloration can be introduced by careless layout. If there are
two parallel screens on opposite sides of the microphone,
a standing-wave pattern will be set up. On the other hand, such
distortion effects may be used constructively, though this is
more often done by placing a microphone fairly close to a single
screen.
For the purposes of multimicrophone music balances, the separation of a large number of individual sources that are physically
close together can be made more effective. Important enough in
mono, this is vital in stereo.
In any studio it helps to have a few screens around. They may
be needed to produce subtle differences of voice quality from
a single acoustic; to separate a light-voiced singer from heavy
accompaniment; or in classical music for backing and reinforcing
horns.
Altering the acoustics
of concert halls
Traditionally, the design of concert halls has been a fallible art rather
than an exact science. In some existing halls of a suitable size it
can be difficult to pick up good sound, and even where careful
experiment has found a reasonable compromise for microphone
placement, many of the audience (whose audible enjoyment may
contribute to the sense of occasion) may still experience poor sound.
Fortunately it is now possible to do something about many such halls
(or studios). Among those that have had their acoustics physically
altered are several of the main concert halls of London.
[4.13] `Flying saucer' diffusers
Royal Albert Hall, London: suspended
convex reflectors of various sizes
break up sound waves that might
otherwise be focused by the interior
of the dome.
Ever since it was built in Victorian times, the Royal Albert Hall
inflicted echoes on large areas of seating and the unfortunate occupants. This was partly due to its size: being very large, for many
places the difference in path length between direct sound and a first
reflection could be 30 m (100 ft), resulting in a delay of nearly 0.1 s,
almost twice as long as required for a distinct echo. This would have
mattered less if it had blended with other, similar reflected sound,
but it did not. The hall was designed as a near-cylinder topped by an
elegant dome. These concave surfaces focused the reflections, and
some coincided with the seating, so the echo was amplified.
Apart from the gallery, where the sound was clean though distant,
there were (as regular concert-goers claimed) only about two places
in the hall that were satisfactory. Staccato playing from percussion
or piano made it even worse. Also, the natural reverberation time of
the hall is long. As a result, the range of music that could be played
to good effect was limited virtually to a single type: romantic music
with long flowing chords. Even this often lacked impact: it was
simply not loud enough, because the hall is too big.
Studios and acoustics
49
Attempts to improve the acoustics without changing the appearance
of the hall failed. The problem was eventually tackled by suspending
flights of `flying saucers' 2±3.75 m (6±12 ft) in diameter: 109 of
these were hung just above gallery level. Taken all together, they
defined a new ceiling, slightly convex towards the audience and
filling about 50% of the roof area.
These polyester-resin-impregnated glass-fibre diffusers (some with
sound-absorbent materials on the upper or lower surfaces) did
largely solve the problem of echo from above (if not from the
oval-sectioned surrounding wall) and at the same time reduced the
reverberation at 500 Hz to more reasonable proportions for a wider
range of music. Even so, the design has been superseded by another,
in which diffusers are also integrated into the ribs of the dome.
In London's post-war Royal Festival Hall the error was in the
opposite direction. Musicians had been asked in advance what
acoustics they would like to have, `tone' or clarity. The answer
came, `tone', but unfortunately what they actually got was a remarkable degree of clarity. The sound was far more dead, particularly in
the bass, than had been expected, largely due to the construction of
too thin a roof shell. Minor measures, such as taking up carpets from
walkways and making surfaces more reflective, had insufficient
effect. While a cough sounded like a close rifle shot, bass reverberation remained obstinately at about 1.4 s. How, short of taking all the
seats out and sending the audience home, could the reverberation
time be lengthened?
The answer here was assisted resonance. Cavity resonators were
used, but not in their normal form: these contained microphones,
tuned to respond to the narrow bands of frequencies present and
re-radiate sound through individual loudspeakers. A hundred channels could each have a separate frequency; allocating each a band of
3 Hz, the whole range of up to about 300 Hz could be boosted. In this
way, resonance was restored at surfaces where, previously, sound
energy had begun to disappear. A reverberation time of 2 s was now
possible, but it could be varied and actually tuned at different
frequencies.
Assisted resonance has since become available as a commercial
package, prolonging a selected range of about 90 frequencies
between 50 and 1250 Hz by half as long again. At the Festival Hall
this dealt effectively with the balanced overall duration of reverberation, but loud brass or timpani could still overwhelm other instruments. Changes to the acoustics have continued.
In both of these venues, however, broadcasters had been able to get
good or fairly good sound even before the alterations by their choice
of microphones and careful placing, so the balance problems (other
than for reverberation time) have not been so severely affected.
Recording companies working with London's many first-rate
orchestras found their own solutions. Audiophiles around the
world may not know it, but the acoustics they have often heard on record
50
Studios and acoustics
are not those of well-known concert halls, but obscure suburban
town halls in places like Walthamstow, Wembley, Watford and
Hammersmith ± narrow, high-ceilinged rectangular boxes that, by
chance rather than musically dedicated design, have much in common with the Concertgebouw in Amsterdam, the Musikvereinsaal in
Vienna and Symphony Hall in Boston, all favourites of musicians as
well as audiences. All these shoebox-shaped halls have two distinctive advantages over their mid-twentieth-century fan-shaped successors. One is simply that the older halls were smaller. Less obvious
was that the solid, narrow, parallel walls deliver a swathe of strong
early reflections, which are now understood to add essential character to musical acoustics.
[4.14] First reflections in a `shoebox'
concert hall
Long parallel side walls create a rich
array of early reflections between
direct sound and reverberation.
Forty years after the Royal Festival Hall, another British hall, the
Symphony Hall in Birmingham, reverted to the size and parallel
sides of the shoeboxes (though with rounded surfaces to add
diffusion from smaller features within the space). But it was also
designed for far greater physical control of its acoustics. It has
concrete-walled reverberation chambers behind the orchestra and
extending high along the sides (adding half as much again to the
volume of the hall), with massive adjustable doors that can be
moved at the touch of a button. There is a 32-tonne counterbalanced
`spaceship' canopy over the orchestra. This can be lowered and the
chamber doors partly closed for smaller forces or a drier sound. To
convert it into a conference hall or for commercial presentations
(and also for loudspeaker-reinforced popular music), thick absorbent
`banners' and a rear curtain emerge, hanging well clear of the
reflective walls for maximum effect.
Built only 30 m (100 ft) from a covered main-line railway, the hall is
mounted on rubber cushions (as also is the rail track), and shielded
from heavy traffic on the opposite side by double skins of concrete.
Ventilation noise is cut by using many large, low-speed ducts.
On stage, a heavy-duty wagon of orchestral risers wheels in, with
built-in wooden sounding boards just for instruments such as cellos
that need them.
Acoustic modelling of studios
[4.15] Controlled concert hall
reverberation
Symphony Hall, Birmingham:
1, Auditorium with parallel side walls.
2, Orchestra with space for organ
behind. 3, Reverberation chamber.
4, Extensions to reverberation
chamber at high level. 5, Massive
doors opening into chamber.
The design of new studios ± or redesign of defective ones ± can
be further refined by modelling their acoustics in advance of
full-scale construction. Modelling reverberation is one of the
fields of audio that has been totally transformed by computers
(see Chapter 12), and studio design has inevitably followed the
same path. It was not among the earliest advances in computer
technology, and the reason for this is obvious as soon as we
consider the computational problem for a reasonably realistic set
of sound reflections.
From each source within the studio, sound travels out in a virtually
infinite number of paths (`rays'), so the first digital compromise is to
Studios and acoustics
51
settle on a practicable distribution of rays. Then the path of each ray
must be traced through each of many reflections in what may be a
very irregularly shaped volume, with the absorption coefficient
known at every point on its inner surface. After a large number of
reflections the tiniest error in this calculation will be gross. Add into
this the approximations required when modelling architectural and
other details, together with the physical compromises required to
allow for diffusion, diffraction and other features of enclosed spaces
that have been described earlier, and the problem begins to appear
intractable.
In fact, the first steps towards successful acoustic modelling were
taken by building and testing physical scale models. In this earlier
analogue world, there was no need for compromise on the number
of rays or number of reflections, but it is obvious that here, too,
there were bound to be enormous compromises. This was done as
follows.
In a model made to a linear scale of one-eighth, sound path lengths
are reduced in the same proportion. With absorbers scaled accordingly, reverberation is also one-eighth of that for a full-size studio.
Surfaces are one-sixty-fourth life size and the volume about onefive-hundredth. Variations in design could therefore be explored at a
cost that was modest compared with that of correcting problems
in the real hall. A special acoustically dead selection of musical
recordings was replayed into the model at eight times normal speed,
and re-recorded with the acoustics of the model added. When this
was replayed at normal speed, the results of various layouts and
treatments could be compared.
The design of a loudspeaker to fill the model with sound at the
required 400 Hz to 100 kHz was inevitably unorthodox. It had three
elements: an 11 cm (41/4 in) thermoplastic cone (for 400±3000 Hz), a
20 mm (8 in) domed plastic diaphragm (for 3±21 kHz), and a cluster
of 45 small electrostatic units mounted on a convex metal hemisphere giving wide-angle high-frequency radiation (at 21±100 kHz).
Microphones to pick up the same raised frequency band ideally
required a tiny diaphragm: the 6 mm (1/4 in) diaphragm used was
rather large for the highest frequencies. It was omnidirectional, so
for a stereo pick-up a spaced pair was used. The (tape) recorder had
to cover 50 Hz to 12.5 kHz at normal speeds and the 400 Hz to
100 kHz band when speeded up.
Structural components and miniature furniture were first acoustically
tested in a small reverberation chamber. Materials with textures
similar to the full-scale object were used where possible; e.g. velvet
for carpets. Perforated absorbers were represented by scaled-down
absorbers of similar construction. An orchestra is widely spread and
mops up much of the sound it creates, so early experiments were
unsuccessful until the musicians, too, were modelled ± in expanded
polystyrene with 3 mm (1/8 in) of felt stuck to their backs. The
biggest problem was the air, which had to absorb eight times more
sound in transit achieved by drying it to 4% relative humidity (compared with around 45% in normal air).
52
Studios and acoustics
The first application of this was to redesign an orchestral studio, a
former ice-rink that had served the BBC Symphony Orchestra for
many years, but which had nagging deficiencies. Many experimental modifications were tried out on a model of it: absorbers
were changed at one end, added under the roof, then covered with
a low-frequency reflective material; the choir rostra were changed,
then switched to a different position, and subsequently replaced
by another design, while various shapes of wooden canopy were
suspended over the orchestra. The results were compared with
each other and (as was possible in this application) with the sound
produced by playing the same test material through the real
studio. The experiment saved more than its cost by just one result:
the big reflective canopy would have been a waste of money.
When built, the chosen design came within 10% of its predicted
performance, an unusually good result in this problem-infested
field.
Scale models that are smaller still (1:50) have also been used, though
plainly with tighter limits on what can be tested. This ratio is too small
for listening tests, but a spark simulates a gunshot and the reverberation can be measured using microphones with a 3 mm (1/8 in)
diaphragm.
Digital alternatives to these physical models have already reached an
astonishing degree of sophistication. Plainly, even an expensive
software program costs much less than a complicated model and is
far more versatile in its application.
The EASE software program is one of several designed to predict
the acoustic performance of large auditoria such as theatres, arenas,
stadiums and churches. Among its many modules are the `tools'
to build a room, with a range of surfaces, audience areas, and
loudspeaker and listener positions. Wall materials can be defined,
together with their absorption and diffusion characteristics.
Loudspeaker data (at least, as supplied by their manufacturers) is
included. Several modules are concerned with the design and
performance of speaker systems and clusters giving, for example, a
3-D view of their coverage, a 2-D analysis of their effect on the room
and a tool for rendering those effects in 3-D. A probe can be used to
investigate the acoustical properties at any given point, and ray
tracing can check for problem reflections (the more complex
program traces up to 19 reflections). An optional EASE `ears'
auralization tool allows users to listen to the calculated response at
any point in the room on their own PC.
Plainly, this is well adapted to the layout of rooms such as cinema
auditoria (theatres), but can also offer many design and analysis
tools to acousticians working on all kinds of halls. With any
kind of acoustic modelling there are substantial savings of
time and money in the preliminary design, and a better chance
of getting good results later. It should be obvious that the
best approach is always to model acoustics first and build the
structure around it, but unfortunately this is not always standard
architectural practice.
Studios and acoustics
53
[4.16] Acoustical control system
(ACS)
Output from microphones above the
performers is fed to a complex digital
system that mixes, delays and feeds
the recombined signals to
loudspeakers (more than shown
here) around the hall. Feedback
between loudspeakers and
microphones is kept as low as
possible. The aim is to simulate the
sound field of a good music studio,
but with variable reverberation time.
Acoustic holography
[4.17] `Acoustic holography'
principles
In this example, the original walls (1)
carry speech rapidly to the back of the
fan-shaped auditorium, so that people
at the rear can hear individual speech
syllables clearly. At the back wall, sound
must be absorbed so that it does not
reduce clarity at the front. 2, Ideal music
auditorium simulated by the acoustical
control system. 3, Simulated sound
path using three of the loudspeakers
lining the walls at the sides and rear.
In practice, there will be many
loudspeakers in a line along these walls.
One further way to create an all-purpose auditorium is to use acoustic holography. Within an existing acoustically rather dry hall, this
aims to construct a range of richer sound fields that are also rewarding to perform and listen in. One such arrangement is called ACS
(Acoustical Control System). The audience sits between rows of
loudspeakers that run along the side and rear walls. The sound wave
from what is in effect a horizontal line source radiates up and down
but does not fan out horizontally to the sides. In consequence,
speakers at the side that are close to the stage radiate only weakly
towards it. Only a little of their sound is picked up by microphones
in the stage area that are in any case set up to discriminate against
sound from the direction of the hall.
ACS employs a matrix of electronic delays and filters to interconnect all microphones and loudspeakers. This reduces acoustic feedback even further to such a safe, low level that it plays virtually no
part in the final sound, and coloration that may accompany it is also
avoided. In the absence of feedback between auditorium and stage,
the speaker system is designed to recreate an ideal wave front in the
audience area. The computer processing that this requires became
practicable in the mid-1980s.
The reverberation time can be increased six-fold to make a small,
relatively dead theatre sound like a cathedral. This allows an auditorium designed for speech to be used for many different kinds
of music, and its acoustics can be instantly reset by the control
computer between pieces, or even between announcements and
music.
Chapter 5
Microphones
A microphone converts sound energy into its electrical analogue ±
and should do this without changing the information content in any
important way. A microphone should therefore do three things:
.
.
.
for normal sound levels it must produce an electrical signal that is
well above its own electrical noise level;
for normal sound levels the signal it produces must be substantially undistorted;
for a particular sound source it should respond almost equally to
all significant audio frequencies present.
In current high-quality microphones the first two objectives are
easily met, although to achieve a signal that can be transmitted
cleanly even along the lead of the microphone may need a transformer or amplifier so close to the head that it becomes part of the
microphone. The third requirement is more modest, and deliberately
so: any claim that sound above the 20 kHz range of human hearing
somehow `feels' better smacks of mysticism and, in fact, the practical limit to broadcast systems may be as low as 15 kHz. But for
a source of restricted range, even less is required of the microphone:
a full audio-frequency range response may pick up unwanted highfrequency noise or catch spill from neighbouring sources that have
a more extended output.
Microphones differ from one another in the way that air movement
is converted into electrical energy. The most important types in
professional use are condenser microphones (also called capacitor
or electrostatic microphones, and including electrets), moving-coil
(often called `dynamic') microphones and, returning to favour after
a period in the wilderness, ribbons. Other principles have also been
employed in the past, including piezoelectric (crystal), carbon,
inductor, moving-iron, ionic and magnetostriction, but these are
not described here: most have practical disadvantages that have
limited their development.
Microphones
55
Nearly all microphones have a diaphragm, a surface that is physically shifted in response to sound waves (although another possible
principle employs a pressure-sensitive semiconductor in an integrated circuit). But there are two fundamentally distinct ways in which
air pressure is converted into movement of the diaphragm. These are
distinguished by differences in their directional characteristics.
Although the output of a microphone may be designed to change with
the angle from a given axis, its frequency response, ideally, should not.
Unfortunately, this ideal is almost impossible to achieve, so it is usual
to define a `useful angle' within which this condition is more or less
met, or to devise ways in which the deficiency is turned to advantage.
Remember, this `electrical analogue' really is analogue. So it is still
subject to the kinds of distortion and noise that afflict analogue
signals. To convert it to `digital' requires a separate, later process.
Microphone properties
The range of properties affecting choice of microphone includes:
Sensitivity. This is a measurement of the strength of the signal that is
produced. In practice, most professional microphones of a particular
type have output voltages within a range of perhaps 10 dB. Taking
moving coils as `average', condenser types can be higher because
they need a head amplifier anyway. Ribbons tend to be lower, unless
they, too, have a head amplifier.
[5.1] Types of microphone
Above: Condenser (electrostatic).
1, Foil diaphragm. 2, Back-plate.
An amplifier is needed so close to the
head as to be considered part of the
microphone. Centre: Moving coil.
3, Thin diaphragm, typically of Mylar.
4, Voice coil fixed to diaphragm.
5, Permanent magnet. Below: Ribbon
microphone. 6, Light, conductive foil
ribbon, corrugated for high flexibility.
7, Permanent magnet, with pole
pieces on either side of ribbon.
8, Transformer (one range has a head
amplifier in place of this). Ribbons,
less popular for some years because
of their limited high-frequency
response and archaic appearance,
have since returned to greater favour.
For both moving coil and ribbon the
magnet still adds weight, and must
also be kept away from magnetic
recordings.
Characteristically, 60 or more dB (further) amplification is required
to reach a level suitable for input to a mixer. The upper limit to
sensitivity is dictated by overload level, beyond which unacceptable
harmonic distortion is generated (or by potential damage). Moving
coils are relatively free of this, unlike condensers, which also
produce more electronic noise (hiss). The limits for both of these
may be described in terms of sound pressure levels (SPL). For
example, a microphone with a stated upper limit of 130 dB is
claiming low distortion up to levels that approach the threshold of
pain, though this may still be less than that close to the mouth of a
loud singer. Moving-coil and condenser types can both have a wide
range, as can some modern ribbons.
Robustness. Moving-coil microphones in particular stand up well to
the relatively rough treatment a microphone may accidentally suffer
when used away from the studio. Condensers and ribbons require
more care.
Sensitivity to handling, e.g. by pop singers. Moving-coil microphones perform well, and condensers with mechanical insulation
between the casing and microphone capsule and electronics are also
good. A reduced low-frequency response helps. Older ribbon microphones are unsuitable, as their diaphragms may have a resonant
frequency of about 40 Hz that is easily excited by any movement
or even a puff of air. Some modern ribbons are engineered to reduce
56
Microphones
[5.2] Microphone sensitivities
These are usually given in dB relative
to 1 volt per pascal. This corresponds
to a (linear) scale in millivolts per
pascal, as follows:
dB rel. to 1 V/Pa
20
25
30
35
40
45
50
55
60
65
mV/Pa
100
56
31.6
17.8
10.0
5.6
3.16
1.78
1.00
0.56
Other units that may be encountered are
dB relative to 1 volt per dyne/cm2: this
gives figures that are 20 dB lower;
1 mV/Pa is the same as 1 mV/10 mbar
or 0.1 mV/mbar.
[5.3] Microphone performance
Figures sometimes quote maximum
sound pressure levels (usually for 1%
harmonic distortion). A conversion to
corresponding linear pressure levels in
pascals is:
SPL (dB)
140
134
128
94
0
handling problems. Cables and connectors should also be checked
for handling noise.
Sensitivity to wind and blasting. Gusting produces erratic pressure
gradients, so microphones that are designed to respond to a pressure
gradient are at a disadvantage. Most ribbons work in this way (see
below), so older designs that have a big, floppy diaphragm are
doubly at risk; modern ribbons are smaller, stiffer and may have
built-in screens. The shape of the microphone head and use of
windshields also need to be considered.
Sensitivity to rumble. Rumble is vibration at 3±25 Hz that can
modulate higher frequencies, making them sound rough. If there is
any danger of rumble, avoid microphones with an extended lowfrequency response ± or use a shock mount or possibly a roll-off filter.
Impedance. An impedance much below 50 is vulnerable to electrical mains hum, so it needs a transformer (or head amplifier) to lift
it. The high (20±50 k
) impedance of a condenser diaphragm might
pick up electrostatic noise from motors or fluorescents, but a head
amplifier prevents that. A moving coil is already in the 150±250 range that allows a cable run of 300 m (1000 ft) or more. Nominally
a microphone should be rated appropriately for the impedance of the
system into which its signal is fed, otherwise the interface between
components reflects some of the signal and this may affect the
frequency response. To avoid this, the impedance of a microphone
should be less than a third of that of the system it is plugged into, and
a ratio of 1:10 is safer. After a transformer or head amplifier (where
required), many microphones have an output impedance of 200 or
less (although some moving coils are as high as 600 ). The preamplifiers built into consoles generally have a 1500±3000 input
impedance; high-quality preamplifiers often have source impedance
matching.
Sensitivity to temperature and humidity variations. Location recording may be subject to extreme climatic conditions, and heat and
humidity can also cause trouble high among the lights in a television
or film studio. If water vapour condenses on to a lightweight
diaphragm, this increases its inertia and temporarily wrecks its
response. A condenser microphone that is transported or stored in
the cold is vulnerable to this, and also to any moisture in its insulation or head amplifier.
Pressure (Pa)
Transient response. Speech and music contain important transients
that are 10±100 ms in duration. High-quality condenser microphones,
with diaphragms less than 0.04 mm (0.0015 in) thick are excellent,
with a response of less than 1 ms, and ribbons are also good. Moving
coils, limited by the mass of diaphragm and voice coil, are less clear.
200
100
50
1
0.000 02
Flexibility of frequency response. Some microphones have a variable polar response, and the switches for this may be located on the
microphone head, or on a separate control unit, or remotely on the
control console. Another switch may control bass response.
Zero SPL approximates to a young
listener's threshold of hearing.
Electronic noise. Microphones generally produce little noise, but as
later stages have improved (especially by the use of digital recording
Microphones
57
media), even a tiny effect becomes more noticeable. In condenser
microphones the head microphone is again a potential problem, for
example during very quiet passages in classical music. Equivalent
noise rating is the electronic counterpart of studio noise. A 20 dB
limit means that the electrical microphone noise is no higher than the
20 dB (sound pressure level) ambient noise of a quiet recording studio.
Power requirements. Moving-coil and (most) ribbon microphones
have no power supply, but condensers need one, both for the charge
on the diaphragm (except for permanently charged electrets) and for
a small head amplifier. Some early condensers had a plug-in power
supply unit. Today, power is usually fed back along the cable from
the console as a `phantom' supply. (But note that any so-called
`digital' microphone also needs a supply to power its internal analogue-to-digital converter.)
Shape, size and weight. Appearance matters in television, and also
size and weight in microphones that are hand-held or attached to
clothing.
Cost. Electrets (permanently polarized electrostatic microphones)
can be cheap, and good moving-coil microphones moderately so;
some high-quality condensers and ribbons are more expensive.
In a given situation any one of these criteria may be dominant,
perhaps even dictating choice of microphone. Manufacturers may
emphasize particular characteristics at the expense of others, and offer
them for some particular application. Popular music attracts a wide
range of different types, some to cope with extreme physical conditions,
others on account of frequency characteristics that are appropriate to
individual instruments or for their `perceived' tonal qualities.
Directional response
[5.4] Polar diagrams
Field patterns of omnidirectional and
bidirectional microphones. The scale
from centre outwards measures
sensitivity as a proportion of the
maximum response, which is taken as
unity. Above: A perfect omnidirectional
response would be the same in all
planes through the microphone (this
diagram is a section through a sphere).
Centre: A perfect bidirectional
response is the same in all planes
through the 0±180 axis. At 60 from
this axis the output is reduced to half,
and at 90 to zero. The response at
the back of the microphone is
opposite in phase. This diagram has
a linear scale, reducing to zero at the
centre, and the figure `eight' is perfect.
Below: The figure-of-eight, shown on
a more conventional decibel scale, is
fatter, but the decibel scale does not
have a zero at the centre.
Microphones fall into several main groups, according to their directional characteristics. Their field patterns are best visualized on a polar
diagram, a type of graph in which the output in different directions is
represented by distance from the centre of the diaphragm.
Omnidirectional microphones. These, ideally, respond equally to
sounds coming from all directions. Basically, they are devices for
measuring the pressure of the air, and converting it into an electrical
signal. The simplest designs of moving-coil and condenser microphone work in this way. The diaphragm is open to the air on one side
only. This type of operation is also commonly called `dynamic'.
Bidirectional microphones. These measure the difference in pressure
(pressure gradient) between the two sides of a diaphragm, and so at
two successive points along the path of a sound wave approaching
from the front or back. If the microphone is placed sideways to the
path of the sound, the pressure is always the same on both sides of
the diaphragm, so no electrical signal is generated. The microphone
is therefore dead to sound from the side and live to the front and rear.
58
Microphones
[5.5] Directivity
The range from pure pressure
operation (omnidirectional) to pure
pressure gradient (bidirectional)
operation. Curve 1 shows the ratio
of sound accepted from the front
and rear. The most unidirectional
response is provided by the
supercardioid (A), for which the
microphone is, essentially, dead on
one side and live on the other. Curve
2 shows discrimination against
ambient sound. With the
hypercardioid response (B),
reverberation is at a minimum and
the forward angle of acceptance is
narrower. In the range between
omnidirectional and cardioid, direct
sound from the front and sides is
favoured, so a microphone of this
type could be placed close to a broad
source. (Note that the close balance
will already discriminate against
reverberation, and that this
microphone will reduce long path
reverberation still further.) Another
measure used is directivity index (DI).
For the omnidirectional response of
360 the DI is zero; for cardioid, with a
solid angle of 131 the DI is 1.7, with
4.8 dB rejection of reverberation;
supercardiod, 115 , DI 1.9, 5.7 dB
rejection; and hypercardioid, 105 ,
DI 2.0 with a 6 dB reduction of
indirect sound.
[5.6] Cardioid response
Left: The sum of omnidirectional and
figure-of-eight pick-up patterns when
the maximum sensitivity of the two
is equal. The front lobe of the `eight'
is in phase with the omnidirectional
response and so adds to it; the back
lobe is out of phase and is
subtracted. Right: The supercardioid
is a more directional version of the
cardioid microphone. It is one of a
continuous range of patterns that
can be obtained by combining
omni- and bidirectional response
in various proportions. As the
bidirectional component increases
further, the rear lobe grows, to
make a cottage-loaf-shaped or
hypercardioid response.
Moving from the front to the side of the microphone, the response
becomes progressively smaller then rises again, and a graph of the
output looks like a figure-of-eight or figure-eight (variations on a
term sometimes used to describe this kind of microphone). The live
angle is generally regarded as being about 100 on each face. For
sound reaching the microphone from the rear, the electrical output is
similar to that for sound at the front, but is exactly opposite in phase.
The simplest designs of ribbon microphone employ this principle.
They respond to the difference in pressure on the two faces of a strip
of aluminium foil. Condenser figure-of-eight microphones have two
diaphragms, one on either side of a charged backplate.
Cardioids (as the name implies) have a heart-shaped response. This
is obtained if the electrical output of a pressure-operated microphone
is added to that of a pressure gradient microphone with a response of
similar strength along its forward axis. The output is doubled at the
front, falls to that of the omnidirectional microphone on its own at
90 to either side and then to zero at the rear, where the two are
antiphase and cancel out. Various ways of combining the two principles are possible: early cardioid microphones actually contained a
ribbon and a moving coil within a single case. There was substantial
pick-up of sound on the front and some round to the side, but
progressively less beyond that.
Microphones
59
Supercardioids and hypercardioids. If the pressure and pressuregradient modes of operation are mixed in varying proportions, a
range of polar diagrams is produced, passing from omnidirectional
through cardioid, then supercardioid to hypercardioid, and finally
bidirectional. The directional qualities within this range may be
described in two ways. One is by the degree of unidirectional
response, indicating the ratio of sound accepted at front and back
of the microphone (i.e. before and behind a plane at right angles to
its directional axis). This is 1:1 for omnidirectional and bidirectional
responses (the two extremes) and maximum, reaching a peak ratio of
13:1, at an intermediate stage. Another attribute of a microphone's
directivity is its discrimination against indirect sound: in practice,
this too is expressed as a ratio, here describing the proportion of the
total solid angle over which the microphone is effectively sensitive
to sound. Again it is 1:1 for omnidirectional microphones; it rises to 3:1
(reverberation is 4.8 dB down) for both cardioid and bidirectional
pick-up, and is higher (a little over 4:1 with 6 dB rejection of reverberation) for a response halfway between cardioid and bidirectional.
The combination of the pressure and pressure-gradient principles in
different proportions is useful for cutting the amount of reverberation received and for rejecting unwanted noises. Some microphones
can be switched to different polar diagrams; others are designed for
one particular pattern of response. When choosing between different
field patterns, note that the breadth of pick-up narrows as the
pressure-gradient component gets stronger. A hypercardioid (also
called cottage loaf) pattern is most evident in the range where the
discrimination against ambient sound is at its greatest; the term
`unidirectional' may be applied to the range between cardioid and
hypercardioid.
Highly directional microphones. These are substantially dead to
sound from the side or rear. They are characterized by their physical
size. A large parabolic dish can be used to concentrate sound at a
microphone placed near its focus, but more commonly a long `interference' tube extends forward from the microphone capsule. For
both, at high frequencies the response is concentrated in a narrow
forward lobe. This degenerates into a broader response at wavelengths that are greater than the major dimension of the microphone.
Even more strongly directional is an `adapted array' microphone in
which a digital signal processor compares the output of five capsules
(one with an interference tube).
[5.7] Highly directional microphone
The unbroken line shows a polar
response for medium and high audio
frequencies. At low frequencies
(broken line) the response
degenerates to that of the
microphone capsule (in this example,
cardioid).
The frequency response of
practical microphones
Practical, high-quality professional microphones do not have a
response that is perfectly flat. In older microphones peaks and
troughs of up to 34 dB were common. This was not just bad workmanship: at one time it was difficult to design a microphone for a
60
Microphones
[5.8] High-frequency effects on a
large diaphragm
The response R at 0 (i.e. on-axis) is
the sum of several components
including: 1, Obstacle effect (which
sets up a standing wave in front of
the diaphragm). 2, Resonance in
cavity in front of the diaphragm.
3, Inertia of the diaphragm as its
mass becomes appreciable
compared with high-frequency
driving forces. 4, This effect can also
be created on a microphone with a
smaller diaphragm by fitting a
ring-baffle around the tip, and can
be increased further by switching
in high-frequency shelving.
level response at all audio frequencies, and even today some deviations may be helpful. In addition, an individual sample may vary by
2 dB or so from the average for its type. Fortunately, few people
notice variations of this order: in a reverberant balance they are
easily overshadowed by variations due to the acoustics. Sometimes
a high-frequency peak is permitted on-axis, to allow for a more level
response at a known angle to it.
Some microphones have a response that falls away in the bass
enough to make them unsuitable for particular orchestral instruments (and certainly for a full orchestra). Much more common is
loss of top, from about 12 kHz on some microphones. Others have a
high-frequency response that goes well beyond this, but with these
the angle at which the microphone is used may be critical: for some,
to obtain the most even response the axis of the diaphragm should
point directly at the sound source. A common defect ± a serious one
for high-quality work ± is an erratic response in the upper-middle
range. Except in individual cases where a continuous, measured
trace of the frequency response is provided with each microphone,
manufacturers tend to publish data and diagrams that somewhat
idealize the performance of their products. This defect may not be
apparent from publicity material.
The size and mass of a microphone (such as a moving coil) affects
its frequency response. In general, the larger the diaphragm and the
larger the case, the more difficult it is to engineer a smooth,
extended top response (but the easier it is to obtain a strong signal).
However, even though smaller-tipped moving coils are available,
there has been a resurgence of interest in the older `peaky' highfrequency models.
For extended, smooth high frequencies, condenser microphones
have particular advantages. The diaphragm has an extremely low
mass, so it responds readily to minute changes of air pressure.
Further, because it must have one stage of amplification near the
microphone capsule (in order to change the impedance before
the signal is sent along the cable), the diaphragm itself can be
small. The precision engineering and head amplifier add to the
cost, but on account of its quality (especially in its frequency
response) and sensitivity, it is widely used in studios. The weakest
link may be the head amplifier: at high volumes, while the capsule
is still within its working limits the amplifier may overload. For
this reason many condensers have switchable attenuation (a `pad')
on the microphone. Always remember to switch it off again after
use.
Microphones
61
[5.9] Ribbon microphone: live arc
This older ribbon microphone has a
ribbon that is 64 mm long by 6.4 mm
wide (21/2 in 1/4 in). Arcs of
approximately 100 at front and back
provide two `live' working areas. The
frequency response at 60 to the axis is
halved, but begins to depart from the
common curve at about 2000 Hz, where
high-frequency effects begin. For a
ribbon that is in a vertical line, those for
variation in the vertical plane (v) are
more pronounced than those for the
horizontal plane (h).
Proximity effect
[5.10] Ribbon microphone: vertical
response
There is partial cancellation of very
short wavelengths (i.e. high frequency)
sounds along the length of the ribbon.
[5.11] Smaller ribbon design
The 25 mm (1 in) ribbon (1) and pole
pieces (2) are enveloped by a spiral of
soft iron (3), which acts as a magnetic
return path. Its smaller size increases
the high-frequency response,
reduces bass tip-up and lowers the
resonant frequency of the foil (to
10±25 Hz, damped further by the very
narrow spacing between foil and pole
pieces). The almost spherical shape
of the magnet system allows simpler
protection from wind effects (4).
A substantial transformer (5) is still
required. The high-frequency
response (6) is extended, but still
rolls off a little (at about 12 kHz).
The pressure-gradient mode of operation, by its very nature, produces a distortion of the frequency response called proximity effect
or bass tip-up: when placed close to a source the microphone
exaggerates its bass. The distance at which this begins to set in
depends on the difference in path length as the signal reaches first
the front then the rear of the diaphragm. Bass tip-up occurs when the
sound source is close enough for the path difference to be a significant proportion of that distance, and is due to the fact that, as it
radiates outward, sound dies away. For spherical waves it diminishes
in proportion to the square of the distance. It is most noticeable on
voices (including song).
All directional microphones that employ a pressure gradient are
affected to some degree. Cardioid and hypercardioid microphones
begin to show this effect, and a pure bidirectional response shows it
62
Microphones
[5.12] Ribbon microphone:
dimensional effects
Left: The sound wave travels farther
to reach the back of the ribbon (1)
than it does to the front. The
effective path difference d is
roughly equivalent to the shortest
distance D round the pole piece (2).
Right: Variation of pressure
gradient (PG) with frequency.
For a large, old design of ribbon the
effective path difference, D, from
front to rear of ribbon is 2.5 cm (1 in).
At this wavelength the pole pieces
create a shadow, and the principle
switches over to pressure
operation.
more. Older (big) ribbon microphones offer the most extreme form
of bass tip-up, as they also require a substantial path difference to
achieve adequate sensitivity. In some cases the effect can be heard as
far out as 50 cm (20 in), becoming increasingly marked as the distance is reduced. In modern, smaller ribbons it is heard at 25 cm
(10 in), and is also heard (though less marked) on cardioids and
hypercardioids at that distance. In all types, with careful use of
distance, it can be used constructively.
[5.13] Proximity effect: variation
with distance
This increases as a directional
microphone approaches a sound
source. For a large bidirectional ribbon
the curves 1±4 may correspond to
distances of 60 to 15 cm (24 to 6 in);
on more compact designs they are
correspondingly shorter.
Bass roll-off can be introduced as equalization (EQ) later in
the chain, but some directional microphones also have built-in
switchable roll-off, which can be set to compensate for the increase
that occurs at some particular working distance. However, any movement back and forth from this chosen position produces changes
in the ratio between bass and middle frequencies that cannot be
compensated by simple control of volume, and to make matters
worse, such movements also introduce changes in the ratio of direct
to indirect sound. If this becomes too intrusive it may be necessary
to revert to an omnidirectional microphone.
At very low frequencies, another effect sets in: the output drops again
because on a very long wave the difference in pressure is small.
When two bidirectional microphones are placed close together a
check should be made that the two are in phase. If they are not,
cancellation of direct sound occurs for sources equidistant from the
two (although normal reverberation is heard) and severe distortion
may be apparent for other positions of the source. The cure is simply
[5.14] Compact ribbon microphone
Much smaller than their forebears,
later-generation ribbons require
precision engineering. In some, an iron
casing is the magnetic flux return path.
The mellow bass and HF presence of
this design make it suitable for pop
music instruments. One version has a
head amplifier in place of the more
usual transformer.
Microphones
63
[5.15] Double ribbon hypercardioid
microphone
Diameter of head: 38 mm (11/2 in). Its
end-fire response (with the band
across the head) can be used as the M
component of an MS pair. With gentle
HF roll-off, and also at LF when set at
1 m or more, it is suitable for piano
balance and several other instruments.
1, Response at 1 m. 2, Bass tip-up at
10 cm. 3, Increased bass at 2 cm.
to turn one of them round (or to reverse its leads). When bidirectional microphones are used to pick up two separate but fairly close
sources, if they can be placed so that each is dead-side-on to the
other source, this allows better control in mixing.
Cardioid and hypercardioid
microphones
[5.16] Condenser microphone with
directional response
For sound from the front, the path
through the delay port is longer, so
there is some pressure gradient
effect. For sound from the back, the
paths to the front and rear of the
diaphragm are the same, so there
will be some cancellation.
[5.17] Cardioid response of highquality condenser microphone
The axial response is substantially
flat to 15 kHz and the response at
90 almost so. The front-to-back
separation is usually poor even in
good microphones of this type.
There are marked irregularities in
the 180 response curve, but they
are at a low level and therefore
generally unimportant.
With true cardioid microphones the useful angle is a broad cone
including at least 120 on the live side ± but which may be taken as
extending to a total included angle of about 180 , although so far
round to the side the overall response is noticeably lower and in some
designs the high-frequency response may be lower still. Continuing
round to the back of the microphone, the output should in theory
simply diminish gradually to nothing, but in practice, though at low
level, it may be uneven, with an erratic frequency response.
The cardioid's coverage is much broader than the figure-of-eight, so
it can be used where space is limited. When working sufficiently
close there is the usual tendency to bass tip-up: in fact, it is half as
much as with a similar figure-of-eight microphone. This reflects the
combination of polar responses that makes a cardioid.
64
Microphones
[5.18] Hypercardioid speech studio
microphone
A moving-coil `workhorse' of modest
cost. With a diameter of 24 mm (1 in),
the low mass diaphragm and coil
allows a good transient response. In a
radio studio several may be used, with
different-coloured windshields for
easy identification, one for each guest
speaker. A good working distance
for speech is 30 cm. 1, Frequency
response at 1 m. 2, Bass tip-up at
10 cm. 3, Tip-up at 2 cm. Also used
for snare drum, toms, hi-hat and
percussion.
The most obvious way of producing a cardioid, by connecting omniand bidirectional units in parallel, is no longer used, as the combination of different types with disparate effects produces an erratic
response.
A better method is to have both front and back of a diaphragm
open to the sound pressure, but to persuade the signal to reach the
back with a change of phase. For this a complicated phase-shifting
network of cavities and ports is engineered. Moving-coil, condenser and even ribbon microphones have all been adapted to this
principle. Systems of this kind can be described mathematically in
terms of acoustic resistances, capacitances and inductances. Using
these, the pressures on both sides of the diaphragm can be calculated for a range of frequencies arriving from different directions,
so that working back from the desired result, for example a cardioid response reasonably independent of frequency, it has been
possible to design appropriate acoustic networks. However, it is
difficult to engineer a really even response in such a complex design.
[5.19] Rugged kick-drum microphone
A cardioid moving coil designed to
withstand high sound pressure levels
with low distortion. For a close balance
on bass drum and bass guitar amps the
frequency response rises at 100 Hz and
there is a presence peak (suitable for
both) at 2±3 kHz. It can also be used for
deep, powerful brass such as tuba and
trombones.
Microphones
[5.20] MS microphone (above)
This microphone combines capsules
employing two different principles in
a single housing. The forward M or
middle microphone is end-fire; behind
this is a bidirectional S or side
microphone. The M response has a
`presence' peak that is suitable for
speech or actuality; that for S is
flatter, but matches the effect of
the peak. Stereo is obtained by
combining the two signals.
65
Switchable microphones
In a switchable condenser microphone one diaphragm is constantly
polarized and the other can have its polarization changed. If the two
diaphragms are polarized in the same sense the microphone operates as
a pressure capsule, as would two single-diaphragm cardioid condenser
microphones placed back-to-back and wired to add outputs. But if the
polarizing current on one diaphragm is decreased to zero (and the centre
plate has suitable perforations) the polar response becomes cardioid.
Taking this a stage further, if the polarizing current is increased in the
opposite sense, so the voltage on the centre plate is intermediate
between those of the two diaphragms, the mode of operation
becomes fully pressure gradient and the output bidirectional. This is
very similar to the symmetrical layout in which a central diaphragm is
balanced between two oppositely polarized perforated plates.
[5.21] Switchable condenser
microphone
Left: The early version of a high-quality
microphone with variable polar response
and, right centre, a later, smaller version
that replaced it. Its side-fire operation is
more obvious. The response is varied by
changing the polarizing voltage to one
diaphragm. 1, Front diaphragm. 2, Rigid
centre plate (perforated). 3, Rear
diaphragm. 4, Multi-position switch and
potentiometer. 5, Polarizing voltage.
6, High resistance. 7, Head amplifier.
8, Microphone output terminals. When
the polarization is switched to position 0,
the voltage on the front and back
diaphragm is the same and above that of
the centre plate: the capsule operates as
an omnidirectional (pressure)
microphone. At B, the capsule measures
pressure gradient and is bidirectional. At
C, the polar response is cardioid. Far
right: Two such capsules in a single
casing, making a coincident pair.
66
Microphones
Even with a high-quality double-diaphragm microphone of this type,
some compromises have to be made. In a particular microphone, the
cardioid response is excellent, the intermediate positions useful and
the omnidirectional condition is good, but the bidirectional condition
has a broad peak in its high-frequency response.
A combination of two types is acceptable for direct middle and side
stereo, and may be favoured for television stereo sound, including
that for television video and film. Unlike an A and B pair, the M and
S capsules do not have to be precisely matched and could even have
different transducer principles, although more often electrostatic
components have been used for both. The `M' component will
generally be cardioid or supercardioid.
[5.22] Soundfield microphone and
its capsule array
The variable polar response is
switched remotely, as are the
azimuth (horizontal response) and the
angle up or down. A control for
`dominance' narrows the response,
giving a zoom effect, within the usual
limits (i.e. there is no highly
directional response available). The
signals from the four capsules can be
recorded separately, and then fed
back through the control box so that
the directional response can be
modified later.
A more exotic combination microphone has four identical condenser
capsules, set close together in a tetrahedral arrangement. Named the
Soundfield, its great flexibility in use allows a wide range of applications. For example, its response can be altered during a live event
(or later, on a four-channel recording from it) to favour a source
from some unexpected direction, which may be outside the plane set
for a standard stereo microphone pair.
Highly directional microphones
For a more highly directional response, a microphone capsule
(which may have any of the characteristics described above) is
placed within an enhanced or focused sound field. The most common
way of achieving this is by the interference method employed in gun
or shotgun microphones, also called line microphones.
Early designs (offering the simplest way of understanding the principle) had many narrow tubes of different lengths forming a bundle,
with the diaphragm enclosed in a chamber at the base. When these
tubes were pointed directly at a subject, the sound pressures passing
Microphones
67
along the different tubes all travelled the same distance, so the tubes
had no significant effect on the sound. But sound that approached
from an oblique angle divided to follow paths of different lengths, so
when the pressures recombined in the cavity before the diaphragm,
differences in phase caused cancellation over a wide range of
frequencies. The further round to the side, the more extreme this
cancellation became, until for sound from the rear the path lengths
varied by up to twice the length of the `barrel' of the `gun'. But there
was, and still is, a limit to effective cancellation and therefore to the
directional response itself. The length of any gun microphone gives
an immediate visual indication of the longest wavelength (and so the
lowest frequency) at which its directional properties are effective.
[5.23] Gun (US: shotgun)
microphone: interference effect
Wave A, approaching from a direction
close to the axis, reaches the
microphone front diaphragm D by
a range of paths that differ only slightly
in length. Cancellation occurs only at
very high frequencies. Wave B
reaches the diaphragm by a much
wider range of path lengths.
Cancellation is severe at middle and
high frequencies. Undelayed sound
reaches the back of the diaphragm via
port C. The polar diagram of the
capsule without the acoustic delay
tube in front is cardioid; with the tube
it becomes highly directional.
[5.24] Gun microphone
1, Housing for head amplifier.
2, Condenser capsule (cardioid
operation). 3, Acoustic interference
tube (with entry ports along upper
surface). 4, Basket windshield for
open-air use (a soft cover may be
slipped over this for additional
protection in high wind). 5, Hand grip.
In today's gun microphones the many tubes have been replaced by a
single narrow barrel, to which the sound has access at many points
along the length. The phase cancellation effect is the same.
Another early, extreme example of a `rifle' microphone had a tube
2 m (6 ft 8 in) long. As an acoustic engineering aid it has been used to
pinpoint the sources of echoes in concert halls. Similar microphones
have in the past been used for picking up voices at a distance: for
example, reporters in a large press conference. But their operation
could be aurally distracting, as it was very obvious when they were
faded up or down, or panned to find a person who was already
speaking. The speech it picked up had a harsh, unpleasant quality
± but was, at least, clearly audible.
Today, commonly-used gun microphones have interference tubes
about 46 cm (18 in) or 25 cm (10 in) long, each facing a unidirectional condenser capsule. The off-axis cancellation is still limited to
shorter wavelengths, with the switchover depending on length of
tube. At wavelengths longer than the tube, the directional response
reverts to the first-order effect of the capsule itself, with little loss of
quality. Gun microphones are widely used by film and television
recordists for location work. They are sensitive to low-frequency
noise from the sides, but if bass-cut can be switched in, traffic
rumble is reduced. They are supplied with a pistol grip that is
attached below the microphone, so they can be pointed by hand.
68
Microphones
A mesh windshield, often wrapped in an additional soft fleece, is
used for outdoor work, but its bulk makes it obtrusive if it gets in
vision. A strong, gusting wind can cause break-up of the signal even
when the windshield is protected by the body of the recordist and
other windbreaks. An alternative is to use `personal' microphones
(described later) but wind can afflict those too ± and the gun microphone offers better quality outdoors.
When used indoors or in any acoustically live space the gun microphone suffers partial loss of its directional qualities because reverberation is random in phase, so is not subject to systematic cancellation.
Indoors, therefore, the response may be no better, nor significantly
worse, than that of the capsule itself. Unless the microphone is
brought close to the subject (which defeats the object of using the
`gun') off-axis sound is likely to be unpleasantly reverberant. In
television or film studios with dead acoustics it can be slung on a
boom and set farther back than a first-order directional microphone,
so it is useful when a wide, clear picture is required. Its discrimination
in favour of high-frequency sound may improve intelligibility.
[5.25] Response of gun microphone
This has a 45 cm (18 in) interference
tube. For high frequencies the cone of
acceptance encloses 30 ; movement
outside this will pass through regions
of marked and changing coloration.
The axial response is reasonably level,
with some roll-off. One version has a
broad peak in the range above 2 kHz to
allow for air losses on distant sources;
this can be switched to a flatter
response for closer sources or where
sibilance is noticeable.
[5.26] Short interference microphone
Overall length about 25 cm (10 in). At
2000 Hz and below, the response is
cardioid. The length of the tube controls
the minimum distance from capsule to
mouth, so that there is little proximity
effect (bass tip-up) even when the tip is
held close to the mouth. Cancellation in
the short interference tube makes the
microphone highly directional only at high
frequencies. It has excellent separation
from surrounding noise when hand-held
by solo singers or reporters, or when
used as a gun microphone for film sound
recording.
The short (25 cm, 10 in) gun can also be used indoors as a hand
microphone. Here it may reintroduce an older problem of directional
microphones, proximity effect, although in this design it is
controlled in part by the distance between mouth and microphone
that is created by the length of the tube, and switching in the
appropriate bass roll-off can compensate the remainder. Some
examples are sold in a kit that permits the use of the capsule either
by itself as a condenser cardioid or with an interference tube. One of
these is based on an inexpensive electret with a tiny head amplifier
powered by a long-life battery.
An earlier method of achieving a highly directional response was the
parabolic reflector. The principle whereby this achieves its special
properties is obvious; equally clearly this, too, degenerates at low
frequencies, for which the sound waves are too long to `see' the dish
effectively (they diffract round the outside instead of being reflected
within it). For a reasonably manoeuvrable reflector of 1 m (39 in)
diameter the directional response is less effective below 1000 Hz.
Also, the microphone is subject to close unwanted noises at low
frequencies unless a bass-cut is used.
Microphones
[5.27] Microphone in parabolic
reflector
The bowl may be of glass fibre or
metal (coated on the outside with
a material to stop it `ringing').
Alternatively, if it is made of
transparent plastics the operator can
see through it to follow action, e.g. at
sports meetings. In another design
the reflector is segmented and can be
folded like a fan for travel and
storage.
69
For picking up faint sounds, the parabolic reflector has one great
advantage: its directional properties rely on acoustic amplification of
the sound before it reaches the microphone, achieving perhaps a
20 dB gain. This offers a far better signal-to-noise ratio than could be
obtained from a microphone that has a similar capsule but employs
phase cancellation. The reflector is therefore ideal for recording
high-pitched birdsong and should be considered for any specialized
purpose where its bulk and poor low-frequency pick-up are acceptable: one sports use has been to pick up contact noise (the clash of
helmets) in American football. It is of less use for the football
(soccer) played elsewhere, as it does little for the lower pitched
thump of boot on ball.
In a reflector, at high frequencies the sound can be concentrated to
near-pinpoint accuracy, or defocused by moving the microphone
along the axis to give a broader pick-up lobe: the latter is often
preferred. An advantage over gun microphones is that reflectors
work just as well indoors as out. Tested in highly reverberant conditions (an underground garage), a parabolic reflector picked up
intelligible speech at twice the distance that could be achieved by
the better of two gun microphones.
To allow for the contribution from the rim, the microphone capsule
should have a cardioid response, pointing into the bowl. But listen for
phase cancellation at wavelengths related to the dish diameter, for
which the reflected signal is weak and out of phase with direct sound.
It is possible to avoid the limitations on frequency response imposed
by the dimensions of the reflector or tube by resorting to more
complex technology: a combination that has an array of four cardioid
electrets plus a fifth operating as a gun microphone projecting from
the centre of the array and a digital signal processor (DSP) to compare
and combine their output. Inevitably the result will be heavier and
more expensive than standard highly directional microphones, but
there is a big jump in performance. The pick-up is far more sharply
focused, with a 20 lobe that with the help of a DSP is available over a
200±2000 Hz working range, which makes it far more versatile than
other highly directional microphones. Curiously, the angle of response
opens out to approach 60 at 4 kHz and above, so at these frequencies
it operates more like a normal gun microphone.
An array of microphones (like an array of loudspeakers) has a
directional response: a row of capsules will cancel signals coming
from the side of the array, or in the case of a plane array, from
directions that are close to this plane. In this `adaptive array' microphone the array forms a diamond pattern that is little more than
50 mm (2 in) across, so its unenhanced directionality would be
limited to very high frequencies ± however, that is not how it works;
instead, it is used as a feed to the nearby DSP to radically modify the
signal from the standard short gun microphone set at the centre of
the diamond pattern. It can also be switched so that only two (the top
and bottom) of the surrounding capsules are used, and with this
arrangement the response opens out to a horizontal lobe of about
20 60 . Obviously, by rotating the array through 90 the lobe
70
Microphones
turns with it, so it is then broader in the vertical plane. If all four
surrounding capsules are taken out of the circuit it reverts to standard
gun microphone operation.
The DSP also has two built-in switchable filters, one to cut extreme
low frequencies, the other to pass a range of 300±5500 Hz. Note that
although the signal is digitally processed, the operation of the
capsules is (as always) analogue, as is the output.
A huge advantage of the adaptive array microphone is its rejection of
off-axis ambient noise (cutting it by about 70 dB) throughout the
mid-frequency range. In a sports stadium, this gives strong discrimination against crowd noise, so it can pick up the thump of boot on
football cleanly, or alternatively the voice of the manager on the
sidelines. In a rodeo it separates out all of the action noises. It also
allows a shore-based microphone following a rowboat race to hear
oars cutting into the water, but virtually eliminates the far louder
chase boat engines.
[5.28] Adaptive array microphone
At about 35 cm (14 in), this is halfway
between the short and long gun
microphones in length, and 7 cm
(23/4 in) across around the capsule array.
Weight: 475 g (163/4 oz) plus almost as
much again for pistol grip shockmount
and windshield (alternatively, it can be
attached to a pole). The digital control
pack and batteries (500 g, 171/2 oz) will
hang on a belt. The mid-range polar
response is much narrower than that
of a gun microphone.
For speech, although most users would prefer a standard microphone
for any normal balance, it can be hand-held for interviews. But the
adaptive array comes into its own when used for full-frequencyrange pick-up at a distance. It can match crowd noise to the actual
section of crowd seen on a zoom lens, or (rather closer) will separate
wanted dialogue from surrounding crowd chatter and noise ± offering a real edge in crucial news operations that must be covered from
a distance.
Noise-cancelling microphones
There are several ways of reducing noise: one is to put the microphone user, a sports reporter perhaps, in a `soundproof' box. But
such enclosures may not keep out the low-frequency components of
crowd noise, and in any case can introduce unpleasant resonances if
they are small. Another technique is to use an omnidirectional
microphone close to the mouth, but in very noisy surroundings this
may still not provide a good balance between voice and background,
and if the noise is still too loud, there is no good way of correcting
the mixture.
An electret cardioid with its response corrected for use close to the
mouth is much better. Some hand-held microphones have a switched
response, so that they can be used either at a standard distance
(60 cm, 2 ft) or very close (e.g. at 6 mm, 1/4 in). Plainly, the user will
need to control the distance in close use with some care, or the
frequency response will be wrong. A head-worn high-quality electret
microphone of this type can be fitted to the individual user, so that
distance (35 mm, 13/8 in) is controlled accurately.
[5.29] Headphone/microphone
combination (out of vision)
Discrimination against noise is
by distance.
In such cases, noise rejection depends on two factors. One is the
extremely close position, which makes the voice much louder than
ambient sound; the other is the microphone's natural bass tip-up
Microphones
71
[5.30] Lip-ribbon microphone
1, The mouth guard is placed
against the upper lip. 2, A stainless
steel mesh acts as a windshield
below the nose. 3, The handgrip
contains a transformer. 4, Case
containing magnet assembly.
5, Yoke of magnet is towards the
mouth, and the ribbon away from
it. The sound paths are marked.
(proximity effect) which, when compensated, reduces middle-to-low
frequencies still further, as there is no bass tip-up on ambient sound
arriving from a distance. Because bidirectional microphones have
the most extreme bass tip-up, they must offer the greatest possible
reduction: with its pressure-gradient operation, a ribbon is subject to
extreme bass tip-up on sound from any very close source.
[5.31] Response of lip-ribbon
microphone
A, Response to spherical wave
(user's voice) at 54 mm (21/8 in).
B, Response to plane wave (ambient
sound), on-axis. This effect is
achieved by equalization, without
which B would be flat and A would
be high at low frequencies, due to
extreme bass tip-up at such a close
working distance.
A design used successfully for many years by sports commentators,
so that they can speak in a normal voice with the roar of the crowd
on all sides, is a lip ribbon. Obviously, the closer to the mouth the
ribbon is, the better ± so long as the explosive sounds in speech and
streams of air from the mouth and nose can be controlled. At a
distance of 54 mm (21/8 in) ± fixed by placing a guard close to the
upper lip ± these problems can be reduced by windshields (which, as
they are so close to the mouth or nose, need to be made of stainless
steel or plastic) and by cupping the ribbon inside the magnet.
Some versions have a variety of equalization settings for different
degrees of bass, but in general the reduction of noise is about 10 dB at
300 Hz, increasing to 20 dB at 100 Hz. High frequencies up to about
7 kHz are all that are needed for speech in these conditions, so above
that the microphone's sensitivity falls away. And, unlike a cardioid,
the figure-of-eight response discriminates against noise from the side.
The lip-ribbon design is not visually attractive because it obscures
the mouth, but has proved effective in almost any conditions of noise
yet encountered in radio, as well as for out-of-vision announcements
[5.32] Compensation for proximity
effect
1, Some microphones have a reduced
bass response that automatically
compensates for the tip-up (2) due to
close working. 3, With the
microphone at a certain distance from
the source, its response to direct
sound is level.
72
Microphones
[5.33] Noise-rejecting cardioid
For use in vision, an electret microphone with built-in pop-shield that eliminates most noise, mostly by extreme close
working, but also by further reduction of ambient low-frequency noise as bass tip-up is compensated. 1, Frequency
response, on-axis, with mouth at 6 mm (1/4 in) from microphone. 2, Frequency response at 0 for sources 60 cm (2 ft) or more
away. 3, Without bass compensation, on-axis at 60 cm (2 ft). 4, Working very close with compensation switched off, there is
a big lift in the bass. 5, Response at rear of microphone at 60 cm (2 ft) or more.
over loud applause in television. For in-vision applications a more
elegant cardioid, either in a hand-held or head-worn design, may be
preferred.
Microphones for use in vision
Everything within a television or film picture, as much as everything
in the sound, adds to the total effect. A microphone in the picture or
a series of pictures for a long time is an important visual element.
Accordingly, many modern microphones have been designed for
good appearance as well as high-quality performance. They may
be silver-grey or black, but always with a matt finish to diffuse the
reflection of studio lights. If the capsule is not intrinsically small it
should be slender, and performers will more readily understand how
they should work to it if any directional properties it has are in line
with the long axis (end-fire), so that in use it points towards them. A
microphone looks better in the edge of frame at the bottom rather
than intruding at the top. Hanging into the picture, it is untidy and
attracts attention.
Table microphones. A pencil shape perhaps 1.5±2 cm (about 3/4 in) in
diameter has room for a directionally sensitive capsule pointing
along the axis; the body of the pencil houses a condenser microphone's head amplifier. If the mounting is to be relatively heavy and
stable and able to protect the microphone from vibration, that cannot
also be made small ± but it can be given clean lines or can be
concealed from the camera in a sunken well in the table.
[5.34] Flexible goose-neck
connectors
These are stiff enough to allow precise
positioning of the microphone.
Microphones on floor stands. For someone standing up, a microphone may be on a full-height stand, which again should be strong,
with a massive base. Many are telescopic. Whatever the design, in a
long shot there is a vertical line very obviously in the picture, and the
Microphones
73
cleaner this is the better. Any clip that holds the microphone should
be neat but easy to release by a performer in vision. Some microphones have only a condenser capsule at face level, plugged on a
thin rigid extension tube that connects it to the head amplifier: the
capacitance of the wiring is part of the design. Some microphone
kits come with a choice of capsules for omnidirectional or cardioid
response and different lengths of column, including a flexible
`goose-neck' section, so that the capsule can be angled precisely
without the need for a separate arm.
[5.35] Spring-loaded clamp
This requires two hands to release
but is more secure than the
easy-release clip.
[5.36] Easy-release clip on floor
stand
This padded clip allows the performer
to take the microphone from the
stand with one hand while in vision.
[5.37] Condenser microphone kit
This is supplied with alternative
capsules for cardioid and
omnidirectional operation. 1, Head
amplifier and capsule. 2, 3, Extension
pieces (these are fitted between head
amplifier and microphone capsule).
4, 5, Windshields. Right, above:
Response curves for omnidirectional
head fitted directly to head amplifier
(i.e. without the extension pieces).
Below: Response obtained when
cardioid head is used in combination
with extension piece.
Hand microphones. These generally have a `stick' shape so they can
be held easily. If its distance from the user's mouth cannot be
predicted or controlled, variable proximity effects (bass tip-up) can
be avoided by choosing one with a response that becomes omnidirectional at low frequencies. An additional advantage of this is that
by its pressure operation at these frequencies the microphone is less
sensitive to wind and breath noise. A hand microphone that becomes
directional at middle or high frequencies should have an end-fire
response (as it is natural to point the microphone toward the mouth)
but the high-frequency pick-up lobe must be broad enough to allow a
good range of angles. If the axis is angled more towards the mouth,
there may be an increase in the high-frequency response. A skilled
user may employ this effect creatively.
Robustness and lack of handling noise are essential, but sensitivity is
less so: the microphone is often held close to the lips, which is
visually intrusive but improves separation from other sound. To
manage sound levels up to 135 dB it may need switched attenuation
(a `pad').
74
Microphones
[5.38] Pressure zone microphone
with flanges
The flanges may be fixed to a stage
surface by tape of a matching colour,
or by double-sided adhesive tape or
adhesive putty. In this case the
response is near-cardioid, and the
response at low frequencies depends
on the distance of the source. Spaced
at intervals of 3±4 m (10±13 ft) along
the apron of a stage, these have been
used successfully for opera. The polar
response (A ± for a source at an angle
of 30 above the stage) is
unidirectional. There are controls to
boost bass (B) and to cut it (C).
[5.39] Pressure zone microphone on
plate
This is easily fixed to a surface by
tape (on top) or double-sided
adhesive (below). The tiny capsule
receives sound through a gap that is
close enough to the surface to avoid
high-frequency roll-off within the
audio range, then broadens out to
allow for a conventional connector.
The hemispherical polar response
does not change with distance, a
benefit when the distance of the
source is unpredictable.
[5.40] Microphone mouse
Polyurethane foam is moulded with
a tubular slot for the insertion of the
microphone, and there is a cavity for
the capsule, which must be very
close to the floor.
Boundary microphones
Generally, microphones are kept away from reflective surfaces
because of the interference fields these produce. However, if the
diaphragm is moved closer to the surface this effect moves up
through the sound frequency spectrum, until at a distance of
1.2 cm (1/2 in) it occurs only in the highest octave of human hearing,
so that on speech and song the effect becomes unimportant. If it
is moved even closer, the full range is restored. It becomes a boundary microphone or, in one commercial range, a Pressure Zone
Microphone (PZM; which sounds better with the American `zee'
pronunciation).
For a sound wave approaching a rigid, solid surface at a right angle,
a microphone close to it registers only pressure variations, not the
pressure gradient, but for a sound travelling lengthways along the
surface it can be measured by either method. A directional microphone is therefore effective only when it is parallel to it. This
suggests an end-fire cardioid microphone with its diaphragm normal
to the surface, directed towards action at a distance. Again, a commercial version is available, but an alternative is to put a standard
microphone in a polyurethane pad with its diaphragm very close
to the surface: this arrangement is called a microphone mouse. To
avoid high-frequency cancellations, the microphone does have to be
very small (or slender).
Microphones
75
Commercial varieties can be engineered to be so flat that the greatest
contribution to their thickness is the padding used to protect them
from floor vibration. Some have pure pressure-operated capsules, so
they are omnidirectional in principle, but in this housing have a
hemispherical response. In all cases, the hard, reflective surface
should, if possible, be clear for a metre or so (several feet) around.
[5.41] Reducing floor noise
To sample the boundary layer close
to a floor, the microphone may be
insulated by a spongy pad and taped
either to the floor or, if this radiates
too much noise, to a board which is
insulated from the floor by a piece
of carpet or underlay, or both.
Floor-mounted boundary microphones are in danger of picking up
noise conducted along or through the floor. Near the edge of a stage,
a simple way of minimizing directly conducted sound is to mount
the microphone on a stand in an orchestra pit, with the capsule on a
flexible swan-neck bending over the edge of the stage and angled
almost to touch the surface. As before, the directional response will
be horizontal. The final limitation will be noise radiated from the
stage surface and travelling through the air to the microphone.
[5.42] Floor-mounted boundary
microphone stereo array
One of many options for the
application of the boundary principle in
stereo, this home-made layout stands
on a 60 cm 60 cm (2 ft 2 ft) base,
with reflectors 30 cm (1 ft) high. The
hinged `V' forms two sides of a 20 cm
(8 in) equilateral triangle.
Personal microphones
A lavalier microphone may be suspended round the neck in the manner
of pendant jewellery ± but rather less obtrusively. In this arrangement it
may also properly be described as a lanyard or neck microphone.
Alternatively, it may simply be called a personal microphone, a term
with more general application, because the small capsules commonly
used are more often simply attached to clothing. For uncompromised
sound quality the microphone will be in vision; for neater appearance it
can be concealed beneath a tie or some other light garment.
[5.43] Electret: permanent
electrostatic charge
1, Metallic foil. 2, Conductive
back-plate. 3, Coating of insulator
with sealed permanent electrostatic
charge. Here the coating is on the
back-plate. It still works as an electret
if the coating is on the foil instead, but
this increases its mass, and therefore
limits high-frequency response.
The microphone should be omnidirectional because, although it is
situated fairly close to the mouth, it is not always at a predictable
distance. Such microphones work reasonably well because the pattern
of radiation from the mouth is also essentially omnidirectional,
although there is some loss of high frequencies reaching the capsule,
particularly if it is hidden in clothing. Some personal microphones have
a raised high-frequency response; this can also be added or adjusted
later in the chain by equalization. Other possible uses for a small unit of
this type, e.g. held in the hand, or hidden in scenery to cover an
otherwise inaccessible position in a set, require a level response.
Personal microphones must be very light in weight as well as being
tiny, but also need to be robust, as they are easily bumped or
76
Microphones
[5.44] Compact personal microphone
Electret capsule, connected by cable to
head amplifier. The microphone can
be used with one of several clip
attachments or a windshield (W),
which very slightly affects the
high-frequency response.
[5.45] Electret circuitry
Compared with earlier electrostatic
(condenser) microphones this is very
simple. 1, Integrated circuit amplifier.
2, Low-voltage d.c. power supply and
resistor. Phantom power may also be
used. 3, Output (cable).
knocked. Some condensers have a separate head amplifier, connected by a short cable: the wire is thin (to be less obtrusive) but
inevitably is vulnerable to damage, and is also a possible source of
conduction noise when rubbed by clothing. Other capsules are
smaller still, with cable moulded to housing. In some, a tiny amplifier is included in the capsule.
Commonly, a tiny electret is used. In this less expensive variation on
the condenser's electrostatic principle, the diaphragm has a charge
sealed within it (`pre-polarized') during manufacture. This eliminates the power supply that would otherwise be required to charge
the condenser, although a small d.c. battery or phantom power
supply is still needed for an amplifier close to the diaphragm. Early
models suffered some loss of high-frequency response as they aged,
so this may need to be checked. Later variants have been designed to
hold their polarizing charge better (often in polytetrafluorethylene),
and the technology (particularly of sealing the surface) has
improved.
[5.46] Versatile, tiny personal
electret
This measures
4.4 mm 7 mm 13 mm
(0.175 in 0.275 in 0.5 in), and is
supplied with a moulded connector
and a range of holders ± clip, tape
down strip, tie-bar and tiepin ± in
black, grey, tan and white, or with a
small windshield or pressure-zone
adaptor. The response is level in the
mid-range, with a smooth, extended
presence peak of 3±4 dB in the
5±15 kHz range. Its relatively large
power supply can be clipped to a
waistband or put in a pocket.
For the electret, a variant on `personal' is its use on musical instruments, to which it may be clipped, often on a small arm or gooseneck arranged to angle it in towards the source. Close to that, the
reduced sensitivity and therefore higher noise level associated with a
smaller diaphragm is acceptable.
In the studio and on location, a personal microphone may be used
either with a cable (heavier beyond the head amplifier) or with a
radio link.
Microphones
77
Radio microphones
In radio, television or film the user of a personal microphone may
need to be completely free of the cable because:
.
.
.
[5.47] Singer's hand-held radio
microphone
The short stub of the UHF antenna is
visually inconspicuous.
[5.48] Communications pouch
A radio microphone transmitter or
talkback receiver can be carried in a
pouch held by a tape in the small of
the back.
in a long shot the cable may be seen to drag ± this is unacceptable
in drama and distracting in other programmes;
in a sequence involving moves, the cable may catch on the
scenery or get tangled, or movement may be restricted to avoid
such mishaps;
the performer may have to move about so much in the intervals
between rehearsals and takes that it is preferable to allow
full freedom of movement rather than keep unplugging the
microphone.
In these cases the cable may be replaced by a radio microphone (also
called a wireless microphone). This has a battery-powered UHF/FM
or sometimes a VHF/FM radio transmitter and aerial. Unless it is
housed within the microphone casing, the transmitter pack should be
small enough to slip into a pocket, be pinned inside clothing or hang
at the waist beneath a jacket. Another option is to place it on a garter,
under loose clothing. If simpler answers fail, put it in the small of the
back, using a cloth pouch of the exact size, with tapes to hold and
support it. (A similar pouch might be used for a television presenter's radio communication equipment.) A VHF aerial (a quarter
wavelength long) may hang free from the waist or, if that would
be in shot, might be drawn upward within outer clothing. It should
be kept as straight as possible, away from the microphone lead,
and clear of the body. A UHF aerial is shorter (only about 8 cm, 3 in)
and can be housed in a stub on the end of a hand-held microphone.
For this it is visually cleaner and easier to handle than a VHF aerial,
but its range may be more confined. Miniature UHF transmitters are
widely (and reliably) used for talk or game shows in the confined
and controlled television studio environment.
Receiving antenna are usually placed where there are likely to be
unobstructed electromagnetic paths from the performer. The equipment can `see' through wood and fabric sets, but not through metal
mesh, and are likely to be affected by other metallic objects such as
camera dollies that move around during a production.
[5.49] Garter transmitter pouch
One of a variety of pouches that are
commercially available.
When setting up radio microphones, after everything else in the
operating environment is in place but before they are handed out,
check that the antennae are suitably placed. This takes two people:
one walks wherever the performer might go, talking over the circuit
to the other, who listens for dead spots. Alternatively, signal strength
should be monitored (it is too low if a light on the receiver goes off).
It may be possible to find a receiving position that works consistently well for any required transmitter position but, in any case,
wherever possible, it is best to use `diversity reception'. This has two
receiving antennae spaced between a quarter and one wavelength
apart, perhaps at either end of the receiver or at a higher point
nearby; the receiver automatically checks and selects the strongest
78
Microphones
signal. This avoids multipath effects, distortion caused by a drop in
level or even total loss of signal when the performer moves though a
position where two radio paths of different lengths are set up, and
interfere. Smaller diversity receivers can be attached to a camera or
mobile recorder.
If several radio microphones are in use at the same time, this can
cause intermodulation signals (audible `birdies'). This can happen if
transmitters get too close together, and is worse if the frequencies
used are close. The cure is usually to change one of the transmission
frequencies ± but it might still be picked up on a third channel.
Occasionally, other equipment in a studio interferes with the radio
link: for example, a signal operating radio-controlled special effects.
Listen also for electrical interference from machinery, vehicles,
mobile phones, television monitors and (especially) disconnected
video cables: if the cause cannot be eliminated, try moving the
receiving antenna, or to discriminate against a distant source, use
directional antennae.
[5.50] Radio microphone and dipole
The microphone contains its own
UHF transmitter and has a stub aerial.
The base of the stick has switches
including on/off, tone and bass roll-off
(for cardioid close working), and LED
condition and warning lights.
In any case, it is safest to anticipate the unexpected with standby
transmitter packs on different frequencies. Obvious frequencies to
avoid are those used for production talkback continuously broadcast
to the studio floor manager and for switched reverse talkback (where
this is used). If it is of a suitable type, the floor manager's radio
microphone is available in an emergency when all others are in use:
it is therefore helpful to standardize, using the same model for both.
As a final safety measure, if possible have a cable version on
standby. For crucial live broadcasts this should be rigged and ready
for use.
In a television studio, existing monitor circuits can be used to carry
the signal from the dipoles to a radio receiver in the sound gallery. To
smooth out unexpected signal variations, the receiver can sometimes
be set for a modest amount of automatic gain (but not in cases where
the performer is going to speak soft and loud for dramatic purposes).
As the transmitter frequency and sensitivity controls have to be pre-set
(where the performer cannot disturb them), the signal can easily be
overloaded, causing distortion, or undermodulated, reducing the signal-to-noise ratio. These can be avoided by introducing compression,
reducing an 80 dB range to 40 dB at the transmitter, which is compensated by expanding it back to 80 dB at the receiver: the whole acts
as a compander. Matched pre- and post-equalization reduces noise.
On the receiver a squelch-mute threshold control is set so that it picks
up the signal when it is available, but mutes when the signal is
switched off.
[5.51] Small diversity receiver
A compact pack that could be used
with a mobile recorder or attached to
a camera. This is far more convenient
than a cable and current designs
should be almost as reliable.
Transmitter power and microphone mute switches will be accessible
to the user, and regular users should be instructed in their use. A
good transmitter should be free from frequency drift and its signal
should not `bump' if it is tapped or jolted. Battery voltage can drop
suddenly when the power source is exhausted, so batteries should
be discarded well within their expected lifetime (which should itself
be reasonably long). Most have automatic early warning of failure;
if not, a log of hours of use can be used to schedule replacement.
Microphones
79
Different wavebands are used in Europe and the USA, so equipment
carried across the Atlantic must be reset (not always easy). The band
used in the UK is allocated to television broadcasting in the USA.
[5.52] A radio microphone pair
Small body pack transmitter and
rack-mounted receiver with diversity
receiver antennae.
Radio microphones can be used in filming (particularly for documentary or news reports), for roving reporters in television outside
broadcasts, by studio programme hosts and for many other purposes.
But even if the quality of the link is high it can still be compromised by
the position of a personal microphone, and despite all the caveats listed
above, clothing rustle is the main hazard. So alternatives should be
considered: in the studio a boom, and on location a gun microphone.
Contact and underwater
microphones
[5.53] Contact microphone
This can be attached to the
soundboard on a guitar or other
stringed instrument. The principle is
electrostatic but is unusual in having a
rubber-coated diaphragm, in order to
give it relatively large reactive mass.
The polar diagram is figure-of-eight.
[5.54] Sound from solids: active mass
1, Contact microphone. 2, Vibrating
surface. 3, The mass of the suspended
magnet holds it relatively still. The
mass and suspension system can be
arranged so that the resonant vibration
of the magnet is at a very low
frequency and heavily damped.
For a source in a fixed position, the
microphone assembly can be mounted
on a rigid arm (4).
Many solids and liquids are excellent conductors of sound ± better
than air. A device that picks up sound vibrations from a solid
material such as the soundboard of an electric guitar is called a
contact microphone. Because in this the whole microphone housing
moves with the vibration of the solid surface, one way of generating
a signal is for a suspended `active mass' to remain still, by its own
inertia. If the moving-coil principle is used, the mass of the magnet
remains almost at rest, while the body of the microphone moves. For
a massive source such as a girder or rail, the weight of the contact
microphone does not matter much. For light sources, including most
musical instruments, the mass of the moving capsule should be low,
in order to avoid damping its motion. Where the condenser principle
is used, the diaphragm requires added mass: in one design, this is
achieved by coating it with rubber.
Where mass is an important part of the design, it is unreasonable to
expect a flat, extended frequency response. A contact microphone
that employs a piezoelectric transducer (e.g. by bending a strip of
barium titanate to generate an electric signal) may emphasize the
mid-frequency range: this may be accepted as part of the character
of the sound or it may be equalized later. Another, an electret, has a
broader response but still suffers a loss of high frequencies (above
4 kHz). This can be compensated by adding an acoustic electret to
80
Microphones
pick up high-frequency sound directly from strings: in the case of the
guitar, it can be fixed underneath the strings, in the air hole, from
which it will receive an additional component.
Contact microphones can be attached with adhesive putty.
To pick up sound under water, one technique is to seal a conventional microphone in a rubber sheath (perhaps a condom). However,
the sound must cross the barrier from water to air, an acoustic
impedance that distorts frequency response and reduces sensitivity
(totally apart from causing embarrassment, hilarity, or both).
[5.55] Miniature hydrophone
The ideal underwater microphone,
9.5 mm (0.37 in) in diameter, with a
rubber compound bonded directly
onto a piezoelectric ceramic. This
permits an excellent sound
transmission from the water, with a
frequency response that extends far
beyond the range of human hearing.
It can be used to capture the
extremely high- or low-frequency
signals used by some marine life
(which we can hear by pitch and time
changing). The response is flat over
the normal audio-frequency range,
in directions at right angles to the
axis. There is a very slight reduction
in some other directions.
It is better that water is separated from the transducer only by a
material such as chloroprene rubber, which is of matched impedance. One miniature hydrophone has a frequency range of 0.1 Hz
to 100 kHz or more, which may seem excessive in terms of human
hearing, but not for that of whales or dolphins. By recording whale
`song' then replaying it at a different speed or putting the signal
through a pitch-changer, the extended frequency range of underwater communication and sound effects can be matched to our
human perception.
Directional characteristics of
A and B stereo pairs
For stereo, an important technique for giving positional information
is the coincident pair (see Chapter 3). If separate A (left) and
B (right) microphones are used they should be matched by manufacture and model (at the very least), because even a minor difference in design may lead to big differences in the structure of the
high-frequency response. An alternative to the matched pair is the
double microphone, in which two capsules are placed in the same
casing, usually with one diaphragm vertically above the other, so
they have a common axis. In one modern paired ribbon design, the
capsules are fixed so that the two figure-of-eight elements are at a
perfect 90 to each other.
In some paired condensers the polar responses can be switched
between figure-of-eight, hypercardioid, supercardioid and cardioid
conditions (as well as omnidirectional, which has no direct use in
this kind of stereo pick-up). The outer capsule can be rotated and the
angle at which it is set is a matter of choice: it is part of the process
of microphone balance. In practice, it may be as narrow as 60±70 ,
but for simplicity in describing how the signals combine and cancel,
consider the case where they are set at 90 , a right angle.
[5.56] Stereo pair in one housing
For convenience the two diaphragms
are usually placed one above the
other, and in this design the outer one
can be swivelled to adjust the angle
between them.
A coincident pair of figure-of-eight elements set at 90 to each other
have their phases the same (positive or negative) in two adjacent
quadrants. The overlap of these allows a useful pick-up angle that is
also about 90 . Outside this angle, the elements are out of phase, so
they tend to cancel out. At the back there is a further angle of 90
Microphones
[5.57] Coincident figure-of-eight
stereo pair
Two ribbon microphones in line, at a
fixed angle of 90 . They have a
smooth, matched response that is
suitable for orchestra, piano or
woodwind. An iron case completes
the magnetic flux return circuit,
helping to make this an unusually
compact design.
[5.58] AB polar diagrams for
coincident pairs of microphones
at 90°
Above: Crossed figure-of-eight.
Centre: Crossed supercardioid.
Below: Crossed cardioid. The MS
diagrams corresponding to these are
shown opposite. Note that although
the subject of crossed microphones
is introduced here in terms of
elements set at 90 , a broader angle
is often better for practical
microphone balances; 120 is
commonly used.
81
[5.59] MS diagrams for the same
coincident pairs
Above: For crossed figure-of-eight
the useful forward angle (W±X) is only
90 . Sound from outside this (i.e. in
the angles W±Z and X±Y) will be out
of phase. The useful angle at the rear
(Y±Z) will (for music) be angled too
high for practical benefit. Centre:
Crossed supercardioid microphones
have a broader useful angle at the
front (W±X). Below: The crossed
cardioid has a 270 usable angle
(between W and X, forward) but only
about 180 is really useful, as to the
rear the response is erratic.
82
Microphones
within which the elements are once again in phase. In the resulting
image, sound sources from the rear of the pair are superimposed on
those at the front (and reversed left-to-right), so that physically
incompatible objects can appear to occupy the same point in space.
[5.60] Stereo pair on bar
This employs two directional
monophonic microphones with their
diaphragms close together. For
nearby voices, it is usually better to
have the capsules side by side rather
than one above the other.
In a hypercardioid pair (crossed cottage-loaf response) the front
lobes must again be in phase. With the diaphragms still at 90 to
each other, the useful pick-up angle is greater, about 130 at the
front. The corresponding 50 at the back is little use for direct pickup, as the frequency response of the individual microphones degenerates at the back. However, if a source to be picked up on a separate
microphone is placed within this angle, at least its position will not
be distorted by the out-of-phase pick-up that would be found at
the side.
Crossed cardioids have a maximum usable angle of up to 270 , but
for direct pick-up the useful angle is only about 180 , as the frequency response of the microphone that is turned away from the
sound source degenerates at angles that are farther round to the side.
As described in Chapter 3, there are two ways of looking at the polar
response of a coincident pair. One uses the usual response curves for
the A and B microphones. The other considers their derived M signal
(obtained by adding A and B) and the S signal (their difference).
The M and S diagrams correspond to the polar responses of another
pair of microphones that can be used to obtain them directly ± a
`middle' (or main) one that has a range of different polar diagrams
and a `side' one which is figure-of-eight. Sometimes A ‡ B and
A B stereo information is taken directly from main and side
microphones, thereby cutting out several items of electrical equipment. A potential problem with this is that the combined frequency
response is at its most erratic at the most important part of the sound
stage, close to the centre ± though the effect gets less as the S
contribution gets smaller. Plainly, the quality of the S component
must be high, and its polar response should not change with frequency. The `M' capsule, which occupies the tip of the casing,
works as an end-fire microphone, while the `S' element, behind it,
is side-fire.
AB pairs are widely used in radio and recording studios; MS microphones may be more practical in television and video recording, as
the `M' signal on its own will be in the same place as the picture.
Windshields
[5.61] Pop-shield
A framework covered with gauze
protects a microphone from gusts of
air caused by plosives, etc., in speech
and song. An improvised version can
be made from a nylon stocking and
coat-hanger wire.
The term windshield (US: windscreen) covers several distinct functions, for which different devices are used.
Some are used to reduce breath effects, particularly plosive `p' and
`b' consonants, in close working. These are sometimes called popshields (US: pop-screens). Where obstruction to sightlines does not
matter, a simple design has silk or gauze stretched over a wire frame
Microphones
83
10 cm (4 in) in diameter: this is fixed between mouth and
microphone.
Some microphones have an integral windshield, typically two layers
of acoustic mesh with a thin layer of a porous material between. So
long as this protects the diaphragm and its cover from the direct
airflow from mouth and nose the shape does not matter. The effect
on frequency response should have been considered as part of the
overall design. Some microphones, designed primarily for use without one, have separate compact metal-gauze shields made to slip on
to them; alternatively, shields made of foamed material will stretch
to fit over a variety of shapes. These may slightly affect highfrequency response, for which EQ may be used if desired. To restore
the nominal frequency response, a shield should be taken off when
the microphone is used at a greater distance.
For a talk studio, foamed shields may be supplied in several colours,
so each speaker can be directed to sit at a clearly identified microphone, and will be reminded by its colour to work to that and
no other.
[5.62] Windshields
These come in a variety of shapes for
different sizes of microphones and
may be made of: 1, plastic foam;
2, foam-filled wire mesh; or 3, mesh
with an outer protective cover and air
space between the mesh and
the microphone.
The other main use for a shield is to cut down wind noise, which
may be partly due to turbulence at sharp edges, or even sharply
curved corners. Ideally, a smooth airflow round the microphone is
required. If the direction of the wind were predictable, the best shape
would be a teardrop, but in practice windshields are usually made
either spherical or a mixture of spherical and cylindrical sections,
typically 10 cm (4 in) in diameter. The framework is generally of
metal or moulded plastic, covered by a fine and acoustically transparent mesh, often of wire and foamed plastic. This should reduce
noise by more than 20 dB.
Mountings
In addition to the equipment already described, there are many other
standard ways of mounting microphones, including a variety of
booms. In concert halls, most microphones may in principle be slung
by their own cable from the roof of the auditorium, but for additional
safety many halls insist on separate slings, and in any case guying
[5.63] Control of microphone
position
1, Tie-cord. 2, Microphone cable.
3, Guy-wire for fine control of position.
84
Microphones
[5.64] Stands and slings
Methods of mounting microphones
(used in radio studios). 1, Suspension
by adjustable wires, etc. 2, Boom
(adjustable angle and length). 3, Floor
stand (with telescopic column).
4, Table stand (but 1, 2 or 5 are better
for table work if there is any risk of
table-tapping by inexperienced
speakers). 5, Floor stand with bent
arm. Most methods of mounting
have shock absorbers or straps.
(by separate tie-cords) may be needed for an accurate final choice of
position and angle. A radio-operated direction control may be added.
Floor stands and table mountings should have provision for shock
insulation, to protect the microphone from conducted sound, vibration or bumps. Where appearance does not matter, this may be a
cradle of elastic supports. In all cases the microphone must be held
securely in its set position and direction.
[5.65] Shock-absorbing microphone
cradle
This is bulky and unsightly but
efficiently guards against all but very
low-frequency vibration. A loop of the
cable should be clipped to the stand.
[5.66] Studio boom
1, The microphone cradle pivots about
its vertical axis, controlled by the angle
of the crank (2). 3, The arm pivots and
tilts, and is extended by winding the
pulley (4). The whole dolly may be
tracked, using the steering controls
(5), and the platform and arm fulcrum
height may be raised and lowered.
6, The seat is used only during long
pauses in action. The operator has a
script rack and talkback microphone
(7). If a floor monitor is difficult to see,
a small personal monitor (8) might be
attached to the boom arm. 9, The
loudspeaker feeds foldback sound to
the studio floor.
A studio boom used in television and film has a telescopic arm that
extends from about 3 to 6 m (10±20 ft). It can swing in almost any
direction and tilt to 45 . The operator stands on the platform to the
right of the boom, with the right hand holding the winding handle
that racks the counterbalanced arm in or out, and with the left on a
crank handle that rotates and tilts the microphone on the axis of its
supporting cradle. The left arm also controls the angle of the boom.
Microphones
85
A seat is available but is used only in long pauses as it restricts
mobility. Some television organizations provide the operator with a
miniature monitor, fixed to the boom arm a little forward of the
pivot. The platform and boom can be raised or lowered together:
they are usually pre-set to a height such that the boom extends
almost horizontally, swinging freely above pedestal cameras. The
whole carriage can be steered, tracked or crabbed (moved sideways)
by a second operator.
[5.67] Articulated fishpole for film or
video
The angle between the two sections
of this boom is fixed with a locking
nut. The microphone hangs just out
of shot without danger of a pole
crossing the corner of the frame.
A hand-held boom, which may be telescopic or have extension
pieces, is called a fishing rod or fishpole. It takes up much less space
and can also be brought in from directions that are inaccessible to
larger booms; for example, it may work close to the floor. In normal
use, it can be supported on a shoulder or anchored beneath the
shoulder to hold it steady. In exceptional circumstances it can be
held high with both arms braced over the head, but directors should
expect this only for short takes. A cranked boom rises at a more
comfortable angle, and then a locked nut controls the angle at which
the extension hangs down to keep the microphone clear of the film
or television picture.
Cables and connectors
[5.68] XLR 3 M
[5.69] XLR 3 F
The figure is the number of pins.
`M' is `male', with pins pointing in the
direction that the signal travels. The
output from a microphone would
normally have `M' pins; the feed to
the control console would go into an
`F' connector. A rugged, heavy-duty
design that is used widely in studios.
[5.70] XLR 3 F
Wall mount with latch.
The most common types of connectors are XLR, DIN, Lemo and
simple jackplugs. The wiring that goes into them falls into two main
categories: balanced and unbalanced lines.
To protect the low-level and therefore vulnerable signal from
many microphones, a minimum of three conductors is generally
employed. Two wires twisted together form a balanced pair that
carries equal and opposite currents. Additional currents that are
accidentally induced in them anywhere along the way should be
in the same direction and can be made to cancel out. Because two
microphones that are used close together must have the same
polarity (to avoid phase cancellation), it is important to identify
each wire in a balanced pair clearly: one is called `hot' and one
`cold' (even though their signals are similar). It is also vital to
ensure everything is wired up correctly; otherwise there will be
an accidental polarity reversal. In addition to the balanced pair
there is an earthed (US: grounded) sleeve that surrounds them and
gives extra protection. This neutral wire is sometimes, confusingly, called `common', a term that properly relates to unbalanced wiring.
An unbalanced pair, used for high-level lines and much domestic
equipment, has a `hot' wire in the middle, and an earthed screen that
is also used to complete the circuit is therefore called `common'. If
an unbalanced pair is wired into a three-pin plug the `cold' and
`common' pins are wired together. There is no problem with polarity: if anything is miswired, it simply will not work. In professional
86
Microphones
work, unbalanced lines are limited to short runs ± rarely more than
3 m (10 ft). This is enough for the wiring between electric instruments and their `amps'.
[5.71] XLR 7 F
[5.72] DIN 7 F
[5.73] XLR 3 wiring
In a balanced cable there are three
wires: 1, Common. 2, Cold. 3, Hot.
For an unbalanced cable (with hot
wire and screen) the common and
cold pins are connected together.
[5.74] Mono jack
6.35 mm (1/4 in).
[5.75] Stereo jack
3.5 mm (1/8 in).
[5.76] Lemo
In some Lemo plugs and sockets the
inner connectors have a half-moon
overlap.
XLR connectors, widely used in studios, are virtually an industry
standard. They are rugged and rather chunky in appearance, especially when compared with some of the delicate professional microphones equipped with them. The number of pins may be indicated
by a suffix: XLR-3, the most common, has three. A cavity in the
base of a microphone contains three projecting (i.e. male) pins: they
point in the direction of signal output, with `cold' off-centre and
between the other two. Wiring is standardized, but unfamiliar equipment should still be checked for polarity. Most consoles have
switches for polarity reversal in case of need, and separate adaptors
are also available for this.
The cable has a female socket at the microphone end and another
male plug to go into the wall socket (or preamplifier). All XLR
connectors have a locating groove to ensure the correct orientation
as the pins are pushed in. Some are secured by a latch that closes
automatically, and released by pressing on the latch as it is pulled
out. Other XLR arrangements are available, including right-angled
plugs, splitters, and other numbers of pins. Non-standard pin directions are also sometimes encountered (more often in connectors
to monitoring loudspeakers), so female-to-female cables may be
required. The XLR system is robust and will withstand accidental
bumps or rough use, but is also rather too massive for applications
where low weight or small size are required.
Another common standard that is used for more compact and also
much domestic equipment is DIN (the German Industry Standard).
The pins are more delicate, and the plugs and sockets smaller. For
connectors that must be secure, a screw collar may be provided.
Amongst others, Lemo connectors (adopting another common standard) are tiny, with fine pins, but tough and reliable for use with
(for example) compact radio microphone transmitters. Most jackplugs (which can be mono or stereo) have either 6.35 mm (1/4 in) or
3.5 mm (a little over 1/8 in) central pins. Take care that jackplugs
attached to microphones are not plugged into headphone or loudspeaker sockets by mistake, and consider replacing them with XLR
connectors.
Three wires (and therefore three pins), one being earthed, are
sufficient for most purposes, including both the signal and the
power plus polarizing voltage for condenser microphones. The
same pins can be used for all of these by raising both of the live
signal wires to a nominal 48 V above ground: this system is called
phantom power, phantom being a term previously used for grouped
telephone circuits that shared a common return path, thereby saving wires. In practice, many modern condenser microphones will
work with any phantom power from 9 to 52 V and some, alternatively, with batteries. The difference in voltage between the pair
and the grounded third wire provides the polarization for condenser
microphones that need it (electrets do not, but still need power for
head amplifiers).
Microphones
87
More complex systems have more wires (and pins) ± for example,
for switching polar response or bass-cut, and also for radio microphones. Old designs of condenser microphones containing valves
(vacuum tubes) ± collectors' items or working replicas that are
treasured by some balancers for their distinctive sound quality ±
require an additional 12 V supply to the heater.
Trailing or hard-wired audio cables will usually have substantial, but
flexible, colour-coded conductors, with braided or lapped shielding
to protect against electrical interference, and with a strong outer
sheath. Somewhat more complex `star-quad' cables have higher
noise rejection: in a studio it will soon become apparent if these
are needed. For longer distances, multichannel cables (with additional coding colours) will terminate in a junction box for the
individual microphone cables that lead through a multi-pin connector to the control console. Fibre-optic cables are not affected by
either electromagnetic or electrostatic fields and can be used for
extremely long runs with virtually no loss, but when used in a
studio require analogue-to-digital conversion at the output of each
microphone.
A radio studio may be hard-wired to offer a choice of two sockets for
each channel, one on each side of the studio. The sockets may be
numbered 1 and 1A, 2 and 2A, and so on, to make the choices clear.
[5.77] Cables
1, Screened pair (good). 2, Star-quad
cable (better). 3, Multichannel cable
`snake' with 12 or 16 pairs, each
having a foil inner shield, within a
single outer sheath.
In television studios, routings are often more complex, as they
may have to go round or through a set (which may include
extensive backings) at suitable points, and also avoid camera
tracking lines and areas of the studio floor that may appear in
vision. So here a very large number of points is provided: some
studios have over 50, some even as many as 100 or more. Mostly
they are distributed around the walls in groups, but some are in
the studio ceiling, so that cables can be `flown' or microphones
suspended from above. A few terminate in the gallery area. It is
convenient to have several multi-way single-socket outputs at the
studio walls so that a single cable can be run out to a terminating
box on the floor, to which, say, 10 microphones can be routed. In
order to minimize induction, sound wiring should be kept away
from power cables (to lighting or other equipment) and should
cross them at right angles. This is also necessary for wiring built
in to the studio structure.
Since microphone cables often carry low-level analogue signals,
they are particularly at risk from earth loops ± also known as ground
loops (one of several technical quality problems described in Chapter 9). If, by mistake, wiring is earthed at two or more points along
its path, this creates a gigantic loop in which electricity can be
generated by any power cable that by chance passes through it.
Since this parasitic current passes along the screening wire, it immediately gets one layer closer to the signal and may be audible as
mains hum. Unscreened wiring carries it directly. This is obvious
when it happens, but may not strike until some item of equipment is
switched on, possibly at just the wrong time. So take care when
laying wiring: it is vital to prevent this `enemy in the camp' from
getting in.
88
Microphones
Microphone checks
When a microphone has been plugged in ± and where more than one
is in use ± a few simple checks need to be made. First, it must be
positively identified at the mixer. Gently tap or scratch the casing
while a second person listens, to be sure which channel carries the
signal. If the console is already in operation, it may have pre-fade
listen (PFL) facilities that can be used so that the main programme
signal is not interrupted. The polar response is confirmed by speaking into the microphone from various sides, identifying them verbally in turn: a dead side will be obvious. For a stereo pair, these
checks ensure that the `A' channel is on the left, and that both are
working. Checks can also be carried out by one person (such as a
recordist on location) by using headphones.
In addition, check microphones for handling noise and decide
whether a pop-shield is needed for close speech. Gently check cable
connections and plugs by hand to see if these make any noise. Cable
runs may need to be covered or taped down, not only to protect them
but also for public safety. Gather up any excess into loops.
A new or unfamiliar microphone may require more exhaustive
tests before use. The manufacturer's literature gives some indication of what to expect, but is not always completely clear. It is
best to start by making a direct comparison with some similar
microphone if one is available; failing that, another by the same
manufacturer (for a common standard of documentation). This
`control test' should show up any obvious deficiency in sensitivity
or performance. Try a variety of sound sources: speech, a musical
instrument and jingling keys (for high frequencies). Check the
effect of any switches ± to change polar response, roll off bass
or reduce output.
A large organization using many microphones may test them more
comprehensively, using specialized equipment in an anechoic chamber ± establishing its own standards, rather than relying on claims
in various manufacturers' literature, which may not be directly
comparable.
Lining up a stereo pair
If coincident microphones do not have the same output level, the
positional information is distorted. A more detailed line-up procedure that can be carried out in a few minutes by two people, one at
the microphones and one to listen at the console, is outlined
below:
1. Check that the monitoring loudspeakers are lined up reasonably
well, e.g. by speaking on a mono microphone (or one element of
the pair) that is fed equally to A and B loudspeakers. The sound
should come from the centre.
Microphones
89
2. If microphones are switchable, for a coincident pair select identical polar diagrams. These may be the ones that will subsequently
be used operationally, or if this is not known, figure-of-eight.
3. Visually set the capsules at about 90 to each other. Check and
identify the left and right capsules in turn. Where microphones
are mounted one above the other, there is no agreed standard on
which capsule should feed which channel.
4. Visually set the pair at 0 to each other, making sure that microphones of symmetrical appearance are not back-to-back (there is
always some identifying point, e.g. a stud or logo on the front, or
mesh at the back of the housing that is coloured black).
5. Listen to A B only, or if there is no control to select this, reverse
the phase of one microphone and listen to A plus reversed B.
6. While one person speaks into the front of the microphones, the
other adjusts for minimum output. If possible, this should be on a
pre-set control for one microphone channel or the other; most
conveniently, it may be a separate microphone balance or `channel balance' control.
7. Restore the loudspeakers to normal stereo output. As the person in
the studio talks and walks in a circle round the microphone, the
reproduced voice should remain approximately midway between
the speakers, whatever the position of the speaker. Except for minor
variations when the speech is from near the common dead axis or
axes of the microphones, if there are any major shifts in image
position as the person speaks from different places there is something wrong with the polar response of one of the microphones; the
test must be started again when the microphone has been corrected.
The earlier part of the line-up procedure is then invalid. As the
person walks round, loudness should vary in a way that coincides
with the polar characteristic chosen.
8. Restore the angle between the capsules to 90 (or whatever angle
is to be used operationally) and again identify left, centre and
right. This completes the test.
Chapter 6
Microphone balance
Earlier chapters have described the physical nature of sound and
how it behaves in an enclosed space, such as a studio. We have
also seen how a microphone converts the energy of sound into its
electrical analogue ± which it should do faithfully (or with changes
that are known, understood and acceptable) and at a level suited
to onward transmission. Now we must put all of these elements
together. This chapter describes the principles that govern the relationship between microphones and their acoustic environment:
sound sources, ambient sound (reverberation) and unwanted sound
(noise). The application of these principles to speech and music are
described in Chapters 7 and 8, and to effects in Chapter 15.
Its surroundings may suggest the way in which a microphone is
used. In a very noisy location, for example close to an aircraft that
is about to take off or among a shouting crowd, the obvious place for
a speech microphone is right by the mouth of the speaker. In these
and many other cases the first criterion must be intelligibility. But
if clarity is possible in more than one limiting position of the
microphone (or with more than one type) there are choices to be
made. In this book the exercise of these choices is called microphone
balance. `Balance' is a broader term than microphone placement,
although placement is certainly a part of balance.
The first object of balance is to pick up the required sound, but at the
same time to discriminate against unwanted noise. Beyond just
finding a place to put a microphone, in professional practice this
also means choosing the best type of microphone, with particular
regard to its directional and frequency characteristics. Another
choice to be made is whether one microphone will cover the sound
source adequately, or if several would do it better. In addition, it is
often possible to choose the position of the source or degree of
spread of several sources, and to select or modify their acoustic
environment.
Microphone balance
91
Where natural acoustics are being used, the second aim of a good
balance is to place each microphone at a distance and angle that
will produce a pleasing degree of reinforcement from the acoustics ± in other words, to find the best ratio of direct to indirect
sound.
Here there is another possibility: natural acoustic reinforcement can
be enhanced later, usually by artificial reverberation (AR) that is
mixed in to the final sound electrically. Taking this further: if the
original, chosen balance discriminates against the natural acoustics
(or a dead environment is used), AR can be added later. Then, for
even greater flexibility, each sound source can have its own treatment. For this reason, and because picking up a sound on a second,
more distant microphone will reintroduce the studio acoustics, we
have as an alternative aim of microphone balance, to separate the
sound sources so that they can be treated individually.
For many kinds of broadcast or recorded sound ± and especially
for popular music ± the term `balance' means even more than all of
that: it also includes the treatment, mixing and control functions that
are integral parts of the creation of the ultimate composite sound.
Stereo microphone balance
[6.1] Compression of scale of width
A sound source that might be spread
over 90 or more in real life (above)
may be compressed on the audio
stage to 60 or less (below).
In stereo (and surround sound), microphone balance has an additional objective: to find a suitable scale of width. Whereas indirect
sound spreads over the entire sound stage, direct sounds should
occupy some clearly defined part of it.
Plainly, a large sound source such as an orchestra should occupy
the whole stereo audio stage and, to appear realistic, must be set
well back by the use of adequate reverberation. On the other hand,
a string quartet with the same reverberation should occupy only
part of the stage, say the centre 20 within the available 60 ;
anything more would make the players appear giants. The small
group can, however, be `brought forward' to occupy more of the
sound stage, provided that the balance has less reverberation. In
either case, the reverberation would occupy the whole available
width.
With coincident microphones, image width can be controlled in
several ways:
.
.
[6.2] Orchestra in correct scale of
width
A broad subject needs enough
reverberation to make it appear a
natural size (1); if the reverberation is
low the subject appears close and
under-size or cramped together (2).
.
The microphones may be moved closer to or farther from the
sound source. Moving farther back narrows the source and also
(as with mono) increases reverberation.
The polar response of the microphone elements may be changed.
Broadening the response from figure-of-eight through hypercardioid out towards cardioid narrows the image. It also reduces the
long-path reverberation that the bidirectional patterns pick up
from behind and permits a closer balance.
The ratio of A + B to A B (the M to S ratio) may be changed
electrically. If S (the A B component) is reduced, the width is
92
Microphone balance
.
reduced. If M (the A + B component) is reduced, the width can be
increased.
The angle between microphone axes may be changed. Narrowing
the angle also brings forward the centre at the expense of the
wings, and discriminates against noise from the sides.
With spaced stereo microphones, their separation and the directions
in which they point have the strongest effect on image width; and the
M/S (A + B/A B) ratio may again be manipulated. With both kinds
of stereo balance, further reverberation may be added from additional distant microphones. This will of course be natural
reverberation, in the same acoustic.
Where individual sources for which no width is required are
balanced on monophonic (or `spotting') microphones, the balance
must be close enough (or the studio sufficiently dead) to avoid
appreciable reverberation, or there will be a `tunnel' effect, extra
reverberation beamed from the same direction. This closeness is also
consistent with application of the `3:1 rule', which proposes that
to reduce spill, adjacent microphones should be over three times
further from each other than from the sound sources they are set
to cover.
[6.3] A quartet in stereo
Above: The quartet uses only a third
of the audio stage, but reverberation
fills the whole width. Below: The
quartet occupies the whole of the
space between the loudspeakers.
Here the reverberation must be light
in order to avoid the players seeming
gross ± such reverberation as there is
appears from behind them.
A position on the sound stage may be established by steering
portions of the signal to the A and B channels (this is also called
panning ± see Chapter 9). If a second mono microphone picks up
the same source unintentionally, it joins with it to form a spaced
pair, which gives a false position; and worse, the amount of
displacement also shifts if the level of one of the microphones
is changed. Close balance, directional characteristics, layout of
sources in the studio and screens may all be used to make sure
this does not happen.
Mono spotting techniques cannot be used for subjects that have
obvious physical size ± such as a group of instruments or singers,
a close piano or a pipe organ. For these, two mono microphones may
be used to form a stereo pair, or an extra coincident pair set in.
The balance test
To get the best possible balance in each new microphone set-up
requires a little experimentation ± and it is convenient to call this
process `a balance test'. But first, think about how you want the endproduct to compare with the original sound. Do you want it to be a
close representation of the original sound? Do you want to adapt it to
suit different listening conditions? Do you want to achieve some
other effect selected from a broad spectrum of modified sound
qualities that might be offered for the listener's pleasure? The endproduct of any balance is never an exact copy, but always an
interpretation. A vital part of the balancer's job is to exercise choice,
making selections from a vast range of possibilities on behalf of the
audience.
Microphone balance
93
`Good sound' is not just a matter of applying rules: its quality is
subjective. But (paradoxically) the balancer's subjectivity must be
allied to a capacity for analytical detachment ± which also means
that, if possible, the results must be judged in the best listening
conditions that any intended audience is likely to use. The surest
way of getting a good balance ± or at any rate the best for the
available studio, microphones and control equipment ± is to carry
out a balance test.
A balance test is not made on every occasion: with increasing
experience of a particular studio and the various combinations of
sources within it, a balancer spends less time on the basics and more
on detail. A newcomer to microphone balance is likely to see a great
deal of attention to detail, and this apparent concentration on barely
perceptible changes may obscure the broader objective, and the way
that is achieved.
Consider first a single microphone (or stereo pair) picking up a
`natural' acoustic balance. A full comparison test will not be needed
very often, but at some stage it is a really good idea to rig at least two
microphones in positions that (according to the general principles
outlined later) should promise an acceptable result. Compare these
two sound balances by using the mixer console and monitoring
loudspeakers to listen to each in turn, switching back and forth at
intervals of 10 seconds or so, considering one detail of the sound
after another until a choice can be made. The poorer balance is then
abandoned and its microphone moved to a position that seeks to
improve on the better. Ideally, the process should be repeated until
no further improvement is possible or time runs out. In practice, this
means advance preparation, creating opportunities to move microphones with minimum disturbance and loss of rehearsal time, or to
select from a wider range that has been pre-rigged.
For a multimicrophone balance the same process may be carried out
for each microphone. Here, what might appear to be a lengthy
process is simplified: there is no need for an acoustic balance
between many sound sources with a single microphone, as this is
achieved in the control desk; nor (in dead music studios) to balance
direct against ambient sound. Where multitrack recording is used
(as in most music recording studios today) some of the balance
decisions can be delayed until the recordings are complete. Mixing
down ± still using comparison tests ± is a part of balance that may be
continued long after the musicians themselves have gone.
For speech, in those cases where the speaker works to the microphone (which includes most radio and some television), a similar
method may be used: place two microphones before the speaker, and
at some stage adjust or replace that which gives the poorer balance,
and so on.
A television balance in which the microphone is permitted to appear
in vision may be similar to those already described. Static microphones are set in to cover musical instruments or to speakers in a
discussion, or a singer may perform to a stand microphone. But
where there is movement and the microphone itself should not be
94
Microphone balance
seen, the process may be reversed: ideally, the microphone is made
mobile (on a boom), working to the performers. With a moving
microphone the balance changes from moment to moment ± ideally
in harmony with the picture. In either case the way it fits either in or
out of vision increases the complexity of any balance problems, to
which the simplest answer may be additional microphones.
In any kind of balance, the main things to listen for are the following:
.
[6.4] Balance test
Compare the sound from
microphones A and B by switching
between the two. The microphones
lie in a complex sound field
comprising many individual
components. For simplicity, the
illustration shows monophonic
microphones. In a mono balance test
listen on each microphone for the
contributions from: 1, Direct sound.
2, Reflected sound, reverberation.
3, Unwanted noise from related
sources. 4, Other wanted sources
and reverberation. 5, Extraneous
noise. In stereo, the principles are the
same, but the sound paths are more
complex. In the stereo balance test,
listen also for scale of width and for
the location of individual components
within the stereo image.
.
.
.
.
The clarity and quality of direct sound. In the case of a musical
instrument this radiates from part of the instrument where the
vibrating element is coupled to the air. It should comprise a full
range of tones from bass to treble, including harmonics, plus any
transients that characterize and define the instrument.
Unwanted direct sounds, e.g. hammer (action) noises on keyboard
instruments, pedal noises, piano-stool squeaks, floor thumps, page
turns, etc. Can these be prevented or discriminated against?
Wanted indirect sound, reverberation. Does this add body, reduce
harshness and blend the sound agreeably, without obscuring
detail or, in the case of speech, intelligibility?
Unwanted indirect sound, severe colorations.
Unwanted extraneous noises.
When trying out a balance, listen to check that bass, middle and top
are all present in their correct proportions, and seem to be in the
same perspective. If they do not, it may be the fault of the studio,
which may emphasize some frequencies more than others, thereby
changing both proportion and apparent perspective. Or it may be
more directly the fault of a balance in which a large instrument is
partly off-microphone (as can happen with a close piano balance).
Remember also that in a live acoustic, a balance that works on one
day may not on the next, when the air humidity is different.
Some variables that may be changed during a balance are:
.
.
.
.
.
distance between microphone and sound source;
height and lateral position;
angle (in the case of directional microphones);
directional characteristics (by switching, or by exchanging the
microphone for a different type);
frequency response (by using a different microphone if an
extended response is required, or otherwise by equalization).
For greater control over these variables, as few of them should be
changed between successive comparisons as is feasible in the time
available. Another possibility for improving a poor balance is to set
in additional microphones to cover any parts of the original sound
that are insufficiently represented.
The balancer working in a studio that has a separate control room to
listen to sound (as most do) can use high-quality loudspeakers to
judge the results. Working outside the studio, these ideal conditions
may not be available, but if artistic quality is both important and
possible, balance tests can still be made, by comparing several trial
recordings on playback. Since the recordist uses these trial runs to
judge how to use the acoustics of the studio to the greatest advantage
Microphone balance
95
(or to least disadvantage, if the reverberation is excessive or highly
coloured), high-quality equipment should be used for playback.
Headphones can be far better than small loudspeakers.
A recordist working on location may be at a particular disadvantage,
one that will not be viewed sympathetically back in the studio. So
skill ± or in the absence of experience, very great care ± is needed.
For reportage, an unchecked balance generally yields results that
have an adequate documentary value. Here, experience is invaluable.
Visual conventions for
microphones
In film and television the normal criteria for placing microphones
for good balance still apply, but in addition there is a second set to
think about. These cover the question of what is visually acceptable.
One important convention concerns the match of sound and picture
perspectives for speech. Plainly, a close picture should have close
sound. Also, if possible, a wide shot should have sound in a more
distant acoustic perspective. However, the need for intelligibility
may override this: a distant group walking and talking across a
crowded room can be made clearly audible as though in a relatively
close perspective, while nearer figures are virtually inaudible; again,
a reporter in long shot may be heard in close-up. In neither case will
the audience complain: they want to hear, and they can.
[6.5] Distance from source
Their directivity (whereby they pick up
less reflected sound) allows some
microphones to be placed further
away from the source (1). Compared
with an omnidirectional microphone
(2), the cardioid or bidirectional
response (3 and 4) can be set back to
1.7 times that distance;
a supercardioid (5) to 1.9 and
hypercardioid (6) twice that distance.
Other conventions concern whether microphones may appear in
vision or not.
For drama, the microphone should always be unseen, as its presence
would destroy the illusion of reality.
For television presentations where suspension of disbelief is not
required of the audience, microphones may be out of vision or
discreetly in vision. In-vision microphones are used where better
or much more convenient sound coverage can be obtained in this
way: because much of television consists of programmes in which
microphones are allowed in vision, they may be seen a great deal
without being noticed. However, their very ubiquity is a point
against them: there is visual relief in getting rid of them if possible.
For film (including films for television), in cases where the illusion
of reality does not have to be fostered, microphones may again
appear in vision, but here the reason needs to be stronger than for
other television, and the use even more discreet. Most film or
location video is shot in places where microphones are not part
of the furniture, so to show one can remind the audience of the
presence of technical equipment (including the camera). Also, in a
film there may be greater variety of pictures or more rapid cutting than
is usual in studio-based television, so a microphone that keeps
popping up is more intrusive.
96
Microphone balance
When a microphone really has to be in vision, the choice depends on
its appearance and sometimes its mobility, as well as on the balance
required. Some appropriate microphone types have been introduced
in Chapter 5, and their use is described in Chapters 7 and 8. But
when microphones must not be seen, the best balance will probably
be obtained with studio booms. Here, too, the equipment has already
been described; and as it is used mainly for speech balance, boom
operation is dealt with primarily under that heading. A television
musical performance presents special problems: its balance should
compare well with that on radio or records, even though television
studios are usually fairly dead. Unless a concert hall is used, this
provides an extra reason for multimicrophone balance.
Multimicrophone balance
In a multimicrophone balance a broad range of techniques is drawn
together, often to include ways of handling reverberation and equalization, of building up a recording by adding track upon track, and
then of mixing down from 16, 24 or even more pre-recorded tracks
to a mono, stereo or surround sound finished product. A few brief,
introductory notes on some of the techniques used are given below.
These go far beyond simple `microphone placement' but may still be
regarded as part of the overall `balance'.
Equalization. Nominally this word means restoring a level frequency
response to a microphone that has inherently or by its positioning an
uneven response. But that is just one way it is used: perhaps even
more often it is introduced as a second stage in the creative distortion
of sound, the response of the microphone in its chosen position being
the first. Equalization controls, whether located in an integrated
dynamics module or assigned to the microphone channel by computer, include midlift (for adding `presence') and bass and top cut or
boost. Many consoles offer several presence bands or the corresponding dip, or a high or low `shelf' in response on each channel.
Artificial reverberation (AR). This augments or replaces the natural
reverberation offered by conventional acoustics. Greater control is
possible, as feeds taken from individual microphones can be set at
different levels. On many consoles an auxiliary feed that can be
taken from one or more channels is fed to a digital AR program,
which may have a great variety of realistic reverberation styles as
well as overall times. Older, far less flexible analogue alternatives
include acoustic `echo' chambers. From whichever is used, the
signal is then returned to the mixer as a separate and usually stereo
source. Sometimes different AR settings are used on different channels. Digital reverberation equipment may also include a range of
other effects.
Compression. Reduction in dynamic range may be used to maintain
a high overall level (in order to ensure the strongest possible
recorded or broadcast signal), or it may have some specialized
Microphone balance
97
application. For example, it might be applied to parts of a backing
sound that without it would have to be held low in order to avoid
sudden peaks blotting out the melody, or to the melody itself, which
might otherwise have to be raised overall and as a result become too
forward. Limiters as well as compressors are used for this.
Together with further stages in the treatment of sound, such as
overdubbing, multitrack recording and then mix-down, whether in
stereo or surround sound, these will all be described in more detail in
later chapters on studio operations. The idea of `balance' may then
sometimes be taken further to include decisions made when one
component of a complex sound is weighed against another, but that
is not the primary meaning of the term.
Note that everything about microphone balance before the later
stages listed above was still `analogue'. `Digital', where appropriate,
begins at the console.
Artistic compatibility: mono,
stereo and surround balances
If an item recorded or broadcast in stereo is likely to be heard by
many people in mono, it is obviously desirable that the mono signal
should be artistically satisfying. There are several reasons why this
may mean modifying the stereo balance (or signal) a little:
Loudness. Whereas a sound in the centre appears equally loud in
both channels, those at the side are split between the M and S
channels. As the S information is missing in mono, side sources
may be lower in volume than is desirable. Sometimes instruments
that have been panned too far to the side sound louder when the
balance is heard in mono. This `centre channel build-up' can be
eased by bringing them back in a little in the stereo balance.
Reverberation. The S signal has a higher proportion than M;
the mono version may therefore be too dry. One way round this is
to make the main stereo balance drier still, and then add more
reverberation on the M channel (if this format is employed). In
crossed figure-of-eight balances, in particular, there is heavy loss
of reverberation due to cancellation of out-of-phase components,
and some people would see this as a reason for preferring crossed
hypercardioids (in which the effect is less) or crossed cardioids.
Aural selectivity. Individual important sound subjects may not be
clearly heard in the mono version, although in stereo they are
distinct because of spatial separation. It may be necessary to boost
such subjects or so place them (towards the centre of the stage) that
in the mono version they are automatically brought forward.
Phase distortion. Spaced microphone techniques may not adapt well
to mono, as the subjects to the side have phase-distortion effects:
systematic partial addition and cancellation of individual frequencies
98
Microphone balance
that no longer vary with position. To check on these, feed A + B to one
loudspeaker.
In addition, there are things that are less of a nuisance in mono than
stereo. For example:
Extraneous noises. Noises that are hardly noticed in mono may be
clearer in stereo because of the spatial separation. Worse, any noise
that moves (e.g. footsteps) is particularly distracting unless it is part
of the action and balanced and controlled to the proper volume,
distance and position. Also, noises from the rear of bidirectional
pairs will appear to come from the front, and movement associated
with them will be directionally reversed.
When surround sound is heard in stereo in the home, it is often
automatically subjected to a process called downmixing. In the simplest version, at the receiver the surround left is added to left and
surround right to right, with centre divided equally left and right. The
subwoofer signal is either blended equally or omitted. If the console
has a `downmix' button, use this to check the effect: the result may be
totally unsatisfactory. If so, consider what might be done to improve
matters. Because of this, a separate stereo mix is often made and
recorded on tracks 7 and 8 of an eight-track mix-down. Super Audio
CD (SACD) and DVD-Audio can carry both surround and stereo and
will reproduce whichever is required. Some programs downmix more
sympathetically (also to stereo headphones).
Chapter 7
Speech balance
Some of the following suggestions may seem almost trivial, but they
all help to make a studio (often, here, a radio studio) run smoothly
and produce good, clean sound recordings or broadcasts.
Speech balance depends on four factors: the voice, the microphone,
and the surrounding acoustic conditions and competing sounds. In a
good balance the voice should be clear and sound natural, perhaps
reinforced and brightened a little by the acoustics of the room that is
used as a studio. There should be no distracting noise, no sound that
is not an appropriate and integral part of the acoustic scenery.
A studio designed mainly for speech and discussion may be about
the size of a fairly large living room, with carpet on the floor and
often treatment on the walls and ceiling to control reverberation.
Listen to both the nature and duration of reverberation: a large, open
studio, even if dead enough, will sound wrong, and a speaker needs
screens set in, not to reduce the studio's reverberation, but to cut its
apparent size by introducing more short-path reflections. A warm
acoustic might match that of a domestic sitting room, but for film
narration audible studio ambience might conflict with picture, so for
this and certain other purposes there is more absorber and much less
reverberation (often, for convenience of layout, in a smaller space).
Microphones for speech
We begin with one of the most basic set-ups in radio: one or two
people at a microphone, and on the other side of the glass, another
to balance and control the sound. (This is of course not the
very simplest set-up: that would have just one person, solo, at a
microphone.)
100
Speech balance
Microphone and acoustics are interrelated. For a given working
distance an omnidirectional microphone requires far deader acoustics than a cardioid or bidirectional microphone, which by their
directivity discriminate against the studio reverberation and any
coloration that goes with it. A hypercardioid, with its greater directivity, reduces these still further. On the other hand, the omnidirectional types (including hypercardioids or cardioids which become
omnidirectional at low frequencies) permit closer working.
[7.1] Microphone position
This shows a good position for
speech, with head well up. The script
is also held up and to the side of the
microphone.
Consider first a microphone for which, by this reckoning, the studio
acoustics are important: the ribbon ± which despite its archaic image
is still used in some studios because of its excellent figure-of-eight
polar response. At its normal working distances, 60±90 cm (2±3 ft),
it has a reasonably flat response throughout the frequency range of
speech. There will be some roll-off above and below that range, but
that is no disadvantage, because these limits discriminate against
low- and high-frequency noise. To study the interplay of voice and
unfamiliar acoustics, experiment with balance tests as described in
Chapter 6. Try different voices: some benefit more by close balance
than others. Do this well ahead of any important recording or broadcast, as a last-minute series of microphone adjustments can be
disconcerting.
To help get a good signal-to-noise ratio, the speaker can be encouraged to speak clearly and not too softly ± if that style is acceptable.
On a bidirectional microphone, two voices may be balanced on
opposite sides: this is easy enough unless they have radically different qualities. An interviewer can sometimes help by adopting a level
that acoustically balances the voices. In an interview where the
primary aim is to bring out the views and personality of the interviewee, it often sounds good if the interviewer's voice is a little
lower in volume ± although the audience may like that more than
some interviewers do.
If the relative loudness of two voices does not sound right on the
balance test, it may help if the microphone is moved slightly, to
favour one or the other. But listen for any tendency for the more
distant voice to sound acoustically much brighter than the other;
rather than allow this, introduce some control (i.e. adjust the gain)
with just a slight dip in level as the louder speaker's mouth opens.
Alternatively, balance the voices separately and closer, perhaps on
hypercardioid microphones. Random variations in loudness are less
tolerable when heard over a loudspeaker than when listening to
people in the same room, so if nothing can be done to stop them
rocking back and forth, or producing big changes of voice level
(without breaking the flow of a good spontaneous discussion), some
degree of control is necessary.
It may help if the microphone is less obtrusive, e.g. mounted in a
well in a table between interviewer and interviewee. Sight lines are
better, but noise from the movement of papers on the table may be
worse.
An omnidirectional microphone at a similar distance would give far
too full an account of the acoustics, so this field pattern requires an
Speech balance
101
even closer balance. A disadvantage is that the effect of movement
in and out on a fixed microphone is more pronounced. However,
with a very close balance the importance of the acoustics is diminished, so compensation by fader alone is less objectionable.
In practice, omnidirectional microphones lie somewhat uncomfortably between those designed to use acoustics and those that eliminate
them. As a result, microphones that really de-emphasize acoustics
are often preferred: it is where the hypercardioids come into their
own. They pick up less ambient sound and allow closer balance ±
partly because of the reduced pressure gradient component (compared with bidirectional types), but perhaps also because they may
become somewhat more omnidirectional at low frequencies. This
last effect may, however, begin to bring a boomy low-frequency
component of the room acoustic back in to the balance, which in
turn is de-emphasized by any bass roll-off in the microphone itself.
Hypercardioids come in a wide range of design characteristics.
Variables include angle of acceptance, also the way in which the
polar response changes at low frequencies, the degree of intrinsic
bass roll-off, and whether additional switched bass cut is provided.
At high frequencies the differentially changing response at angles of
30±45 to the axis may be used for fine-tuning the response to
particular voices. For increased intelligibility, especially against
competing background noise, use a microphone (or equalization,
EQ) with a response that is enhanced somewhere between 3 and
10 kHz, but check for sibilance.
Using hypercardioids, voices can be close balanced individually or
(for greater numbers) in close pairs, and can sound acoustically dead
even in a live room, without sounding harsh. This style is also suited
both to voice-over on film and to radio documentary narrative, as
well as news reporting and links between segments with different
qualities and acoustics. But it does also emphasize mechanical
noises from the lips and teeth that sometimes accompany speech,
and increases the danger of blasting (popping), particularly on plosive `b' and `p' sounds.
[7.2] Popping
Problems with plosive `p' and `b'
sounds. 1, Plosives spoken (or sung)
directly into a microphone may send
the diaphragm beyond its normal
range. 2, Speaking on-axis but to the
side of the microphone may cure this
without changing frequency
response. 3, Speaking across and
past the microphone may cure
popping and also change frequency
response. 4, Speaking across the
microphone but still towards it may
still cause popping. 5, A simple
screen of silk or nylon mesh over a
wire obstructs the puff of air, but
does not significantly affect the
sound.
There are several ways to deal with popping. To sample the cause,
project `p' sounds at your hand: you will feel the air blast that might
push a microphone diaphragm beyond its normal limits. Then move
the hand a little to one side. This demonstrates one solution: to speak
past the microphone, not directly into it. But a windshield (popshield) may also be needed, if not already part of the microphone
design. A screen can be made up (as it was originally) from coathanger wire and borrowed tights. Set it in about 10 cm (4 in) from the
microphone.
Some disc jockeys prefer a close, intimate style with very little of the
studio acoustic in the sound. It is possible that the wish to work close
to the microphone is partly also psychological: the performer gets a
feeling of speaking softly into the audience's ear. A male disc jockey
with a soft, low voice only makes it worse by gradually creeping in
to an uncorrected directional microphone and accentuating the bass
in his voice, particularly when presenting pop records that have what
102
Speech balance
in other fields of recording would be regarded as an excess of top.
Once a good balance has been found, microphone discipline is
necessary.
Interviewing with a hand-held
microphone
For an interview using a hand-held omnidirectional, cardioid or
hypercardioid microphone, experiment with four basic balances:
.
.
.
[7.3] Interviewing techniques
Position 1: Microphone at waist or
chest height ± sound quality poor.
Position 2: At shoulder or mouth level
± sound quality fair, or good if
interviewer and interviewee are close
together in satisfactory acoustics
and quiet surroundings.
Position 3: Moving the microphone
from one speaker to the other ±
sound quality good. In a fourth
variant, the microphone favours the
interviewee most of the time.
.
Hold the microphone at or just above waist level, so its presence
is unobtrusive. This has sometimes been tried, either to keep it
out of sight on television or when it is felt that the interviewee
might be put off by anything closer. It gives poor quality (very
poor, if the microphone is omnidirectional), with too much
background sound or too reverberant an acoustic. But a picture
can distract conscious attention from sound quality, so provided
speech is clearly intelligible this balance may sometimes
be permitted. The cone of acceptance should be as narrow as
possible.
Hold it between the two speakers at chest or neck level, with the
face of the microphone pointing upward. If it has cardioid to
hypercardioid response, both speakers must be within the angle
of acceptance, and this may be adjusted during dialogue to favour
each in turn. For television, this balance is better.
Moving the microphone close to each mouth in turn. This balance
is needed particularly with omnidirectional microphones, for
which the distance is judged according to the acoustics and the
noise (unwanted) or `atmosphere' (wanted) in the background.
The interviewee may find the repeated thrust of the microphone
distracting, but if the interviewer times the moves well the overall
sound quality will be good, and any wind-noise and rumble is less
than with directional microphones that are held further away.
Hold it close to and favouring the interviewee, with only slight
angling back during questions, the interviewer relying more
on clarity of speech, voice quality and well-judged relative
volume.
Where a monophonic balance is heard as a component of stereo, try
to discriminate against background sound and reverberation, so
those do not seem to come from the same point. With a stereo
end-fire interview microphone, check the directional response. This
may be clearly indicated on the microphone, but if not, as a reminder, you might mark it with an arrow on tape on the side of the
microphone.
Because the interviewer should be close to the subject, this may feel
less confrontational if they stand at about 90 or sit side by side. If
the loudest extraneous noises come from the direction they are
facing, the microphone can be angled to discriminate against it.
Speech balance
103
Three or more voices
[7.4] Presenter and guests: I
All microphones are hypercardioid,
arranged to allow individual control of
each voice. 1, In this layout for radio,
the presenter has a condenser
microphone with a moving coil as
stand-by. 2, Moving-coil microphones
for guests, who take their places at
the table while records or
pre-recordings are played. 3, For an
additional guest, if required.
4, Headphone feed points are under
the edge of the table, with switches
to select the feed.
[7.5] Presenter and guests: II
Another table layout for radio.
[7.6] Acoustic table with well
The table top has three layers:
1, Perforated steel sheet. 2, Felt.
3, Woven acoustic covering (of the
kind that is used on the front of
loudspeakers). These are not
bonded together, but held at the
edges between wooden battens.
4, Wooden microphone platform,
suspended at the corners by
rubber straps (5).
For three or four people in a radio interview or studio discussion, the
ribbon microphone has been widely used in the past, and still has its
advocates. With its bidirectional response a round table layout is not
possible: four speakers must be paired off, two on each side, or three
divided two and one. The microphone can be moved or angled a
little to favour lighter voices, and two speakers on the same side of
the microphone should sit close together so neither gets off-microphone. Participants should be placed so those who are side by side
are more likely to want to talk to the others across the table. The
change in quality when one turns to a neighbour can be marked, and
leaning back at the same time magnifies it.
A more common arrangement has two to six people around a table,
each on a separate hypercardioid microphone at about 30 cm (1 ft).
The ambient sound pick-up is that of all the open microphones
combined, and in total is broader than that of a single cardioid or
bidirectional microphone, but the much closer balance more than
compensates for this, and in any case allows greater individual
volume control. If the order of speakers is known (or can be anticipated by watching the discussion), some faders can be held down a
little, favouring those already talking. Different-coloured foam covers on the microphones are used both to guide speakers to the correct
place to sit and, by colour coding the fader keys, to match these in
turn to the voices. With more people, one microphone may be
assigned to each two speakers, who must be close together.
Any nearby reflecting surface affects a balance: the reflected wave
interferes with direct sound, producing phase cancellation effects
and making the quality harsher. To reduce this, tables in radio
studios (and for some television discussions too) can be constructed
with a surface of acoustically transparent mesh. If a microphone well
is set into this, perhaps with a small platform suspended by springs
or rubber straps, it should be open at the sides. For insulation against
bumps or table taps, rubber is better when in tension than it is in
compression.
In a sufficiently dead acoustic there is one further possibility: an
omnidirectional boundary microphone on a hard-topped table.
104
Speech balance
Studio noise
[7.7] Round table discussion
Six speakers with a cardioid
microphone, either suspended above
them (1), or sunk into a well (2). This
is not ideal, as all speakers are
somewhat off-microphone and are
likely to be a little too far back.
For many purposes (e.g. narration, plays, etc.) background noises are
undesirable. They are reduced to a minimum by good studio discipline, the elimination of potential problems during rehearsal, and
critical monitoring. A check at the time of the initial balance will
show if room atmosphere is obtrusive when compared with voice.
`Atmosphere' means the ambient noise that pervades even quiet
rooms, and if it seems loud and cannot be reduced, then either voice
levels might be raised or the balance tightened. Unfortunately, the
sort of voice quality that stands out really effectively against
background noise is more suitable for a hard-sell commercial than
ordinary talk.
Even in radio studios designed to discriminate against external
noise, there can be problems: ventilation systems may carry noise
from other parts of the building, or the flow of air can sound noisy
when compared with quiet speech. Structure-borne noise is almost
impossible to eliminate, and rebuilding work is complicated by the
need to avoid all noisy work when nearby studios are recording or
on air.
Noise may also come from the person speaking. A lightly tapped
pencil may resound like rifle fire. A slightly creaky chair or
table may sound as if it is about to fall apart. Lighting a
cigarette can be alarming; lighting several, a pyrotechnic display.
Tracking fainter sounds to their source may prove difficult ±
embarrassing even. A loose denture may click; a flexed shoe
may creak. Some people develop nervous habits at the microphone:
for example, a retractable ballpoint pen can become a menace
as unseen hands fiddle with it, to produce sharp, unidentified,
intermittent clicks.
Several things can make such noises even more irritating. One is that
some people listen to broadcast or recorded speech at a level that is
louder than real life, so a quiet voice is reproduced at much the same
volume as one that shouts. In consequence, extraneous noises
become louder too ± and, worse, if heard monophonically, appear
to come from the same point in space as the voice. In contrast to a
visually identified sound that is seen in context and so is accepted,
unidentified noises do not. Many experienced broadcasters know
this and take it into account. A studio host will notice that a guest's
movements are not silent and say, `Now, coming to join me, is . . . '
When explained, a noise is reduced to its proper proportions, so is
not consciously noticed by the listener.
[7.8] Two bidirectional
microphones in mono
Voices A and B are on the dead side
of microphone 2 and vice versa: this
is a simple balance for a game show,
etc., recorded before an audience.
Any voice at E would be picked up on
both ± provided the two were in
phase (as they should be).
Handling a script
A frequent source of extraneous noise at the microphone ± and an
irritating one ± is paper rustle. Ahead of any broadcast or recording
Speech balance
105
from a script or notes it makes sense to give an inexperienced
speaker some suggestions on how to handle paper quietly:
.
.
[7.9] Avoiding script noise
Corners of script turned up to make it
easier to lift pages noiselessly to one
side.
.
.
.
.
[7.10] Script rack
No reflections from the script can
reach the microphone. The rack may
be made of acoustically transparent
mesh. The speaker might need to be
reminded not to look down when
reading from the foot of the page.
Use a fairly stiff paper (copy paper is good). Avoid using sheets
that have been folded or bent in a pocket. Avoid thin paper (e.g.
airmail paper or flimsies).
At a table, take the script apart by removing any paper clip or
staple. It is not easy to turn pages silently, so prepare a script by
bending up two corners, so each page can be lifted gently aside.
(Sliding one page over another will be audible.) Alternatively,
hold the script up and to the side of a directional microphone.
To avoid reflections of sound from the paper that will interfere
with direct sound and cause coloration, hold or place the script a
little to one side. But do not turn off-microphone or drop the head
to read it.
When standing at a microphone (as actors who need to change
perspective must do), leave the script clipped together. Hold it well
off-microphone when turning pages ± or when using a bidirectional microphone, hold the script up and to the dead side of it.
When there are several people in a studio (e.g. to perform in radio
drama, a sketch or a complex commercial) they should not all
turn their pages together. If performers remember to stagger the
turnover, they will also remember to turn them quietly.
If there are only a few sheets of script, try putting them in clear,
stiff plastic envelopes.
One device sometimes used is a script rack. Sloping up at an angle to
the microphone, it sends reflections that might cause coloration
away from it. Unfortunately, it does not prevent a persistent offender
from looking down to read the bottom of each page.
Speech in stereo
Whether intended for radio, television or even film, much speech
that will form part of a stereo image is, initially, balanced on
monophonic microphones. In particular, radio announcements are
normally taken mono, and are often placed in the centre, so that
listeners can use this to balance two loudspeakers. If the announcement is in a concert hall, a monophonic microphone is again used
with a relatively close and dead balance, perhaps with a stereo pair
left open at a distance to add stereophonic reverberation. The
announcer and microphone are placed where little direct sound can
be picked up in stereo; otherwise the position of the announcer might
shift (with a change in voice quality) when stereo microphones are
faded up or down.
In discussions, individuals are panned to a range of positions on an arc.
One reason for avoiding a stereo pair on speech balances for which no
movement is expected is that any slight lateral movement close to the
microphone is emphasized and becomes gross on the sound stage.
106
Speech balance
Boom operation
In contrast to radio, where the microphone is usually static, and the
performer works to it, in television there are cameras, lights and
action, and the microphone must work to the performer. Radio and
television differ most strongly when, in addition, the microphone
must be out of the television picture, as the performer moves freely
within it. The `microphone with intelligence' that moves and turns
with the action, always seeking the best practicable balance, is the
boom microphone. Although its greatest virtue ± a capacity to cover
long, continuous takes of multi-camera action ± is less in demand
now that digital editing makes it easier to work in television in
shorter sequences or even shot by shot, boom operation is still
central to the subject of speech balance in television, and film too.
Almost everything a boom operator does is related to or affects
balance, so boom microphone coverage is introduced here. How
boom and other television studio operations are planned is described
in Chapter 20.
From a given position a boom can cover a substantial area, and the
platform can be tracked to follow extended action, or repositioned
between scenes. In an ideal balance for best match of sound to
picture, the microphone is above the line between subject and
camera. One factor that affects its main, chosen position is the
direction the key light comes from: the boom will generally angle
in from the opposite side. Another factor is ease of movement: it is,
for example, quicker to swing the arm than to rack in and out when
shifting from one subject to another. A boom operator can quickly
turn the microphone to favour sound from any chosen direction, or
(almost as important) to discriminate against any other. Characteristically, in mono, or for the middle (M) signal in stereo the microphone will be cardioid or near-cardioid, with a diaphragm that is not
noticeably affected by rapid movement and is also insensitive to or
protected from the airflow this causes, so that the operator is not
restricted in speed of response to action. In practice it is best to use
a directional condenser or perhaps moving-coil microphone, with
a shock mounting and windshield (although the most effective windshield is bulky and can make movement clumsy). For stereo pick-up,
MS microphones are often preferred: the `M' component will generally be located cleanly within the picture.
The boom operator has a script (where one is available) and generally moves the microphone to present its axis to favour whoever is
speaking or expected to speak. Although a cardioid has a good
response over a wide range of directions, there may be a 2 or 3 dB
high-frequency loss at about 45 to the axis, increasing to perhaps
5 dB at 60 , so for the best coverage all these twists and turns really
are necessary. The microphone will be dropped as close to performers as the picture allows and this closeness accentuates any angle
there may be between speakers. This means that by proper angling
there can be not only the best signal-to-noise ratio for the person
speaking, but also automatic discrimination against the one who is
Speech balance
107
not. If both speak together, it may be decided to favour the person
with the most important lines, or perhaps to cover both equally. With
performers of different head heights the microphone should be placed
either well in front of them or to the side that favours the shorter.
The microphone position must be a compromise between the positions of all the speakers, and if they move apart, or one turns away,
the split can become impossible to cover by a single boom. In this
case a second microphone may be set in on a stand out of vision, or
hung over the awkward area, or concealed within the picture. Sometimes a second boom is used, but this increases lighting problems.
For `exterior' scenes that are shot in a studio, the microphone must
be kept as close as the limits of the picture allow. To contrast with
this, for interiors use a slightly more distant balance or add reverberation.
Because a number of different types of microphone may be used for
a single voice, the control console in a studio has equalization
associated with each channel that can be used to make the microphones match as closely as possible. If correction is needed, the boom
microphone is usually taken as the standard and the others matched
to it before any further treatment (such as `presence') is applied.
[7.11] Boom coverage of a large
group on TV
Some groups are difficult to cover on
a single boom. In these examples,
two cardioids are used. Above: Boom
A covers speakers 4±8; B covers 1±5.
Below: Boom A covers speakers 5±9
(with 1 and 4 at low level); B covers
1±5 (with 6 and 9 at low level). For
other examples of boom coverage,
see Chapter 20.
Another justification for the boom is that given suitable (rather dead)
acoustics, it automatically gives a perspective appropriate to the
picture: close shots have close sound; group shots have more distant
sound. The boom operator uses vision monitors in the studio to
check the edge of frame, dropping the microphone into the picture
from time to time during rehearsal. A sight line against the far
backing can be used to set and check its position for recording or
transmission.
In television the sound supervisor can call the boom operator (and
other staff on the studio floor) over his own circuit, which cuts in on
the director's circuit to the operator. (This may be automatically
preceded by a tone-pip to indicate that the message is from the sound
supervisor.) During rehearsal and any time when studio sound is
faded out, the boom operator can reply or call the supervisor on a
separate reverse talkback circuit. A boom therefore has a multicore
cable, permitting communication as well as transmitting programme
sound, and returning the mixed sound to the operator's headphones.
In some kinds of television discussion the frame is unpredictable,
but this is partly compensated by the influence of an audience, which
limits how far performers will turn. For this, the boom may be
equipped with a highly directional microphone. Although the inertia
of a tube that is nearly half a metre (18 in) long makes it clumsy to
handle for big, rapid movements, it can be placed farther back.
Where two people are in a `face-to-face' confrontation, but far
enough apart for a boom operator to have to turn a cardioid microphone sharply to cover each in turn, two hypercardioids may be
better, pointing outwards at about 90 to each other for low close
working. In a drama this would restrict the use of the boom for other
scenes, so in a similar situation ± or, for example, for a number of
people round a large table ± two booms might be brought in.
108
Speech balance
A boom can cover action on rostra built up to a height of, say, 1.8 m
(about 6 ft) provided the performers keep within about 3 m (10 ft) of
the front of the high area. The boom platform should be positioned
well back, to keep the angle of the arm as low as possible.
The boom operator should occasionally take off headphones to
check for talkback breaking through or for other sound problems
in the studio and report them back to the supervisor. In addition,
being visible to artists, the operator should react positively to their
performance.
Slung and hidden microphones
In television balance one aim is to keep the number of microphones
to a practicable minimum, and the use of booms helps towards this.
But unless set design and action are restricted there may be places
that are inaccessible to booms. For these, additional microphones
may be slung above the action ± ideally, again, just out of frame, and
downstage of the performer's planned position, so that sound and
picture match. If a slung microphone has two or three suspension
wires, its position can be adjusted in rehearsal, and if there is any
danger of its shadow falling in shot, provision should be made for it
to be flown (raised out of shot) when not in use.
Microphones are sometimes concealed in special props shaped over
wire gauze, then painted or otherwise disguised, though an electret is
small enough to be hidden more easily. Placed on or very close to a
reflecting surface, it acts as a boundary microphone, but check for
the effect of hard surfaces further away that may modify the top
response. Sometimes equalization can help by reducing highfrequency peaks so the quality is not so obviously different from
that of the boom, but it cannot restore cancellation.
Microphones in vision
The presence of a microphone in vision must be natural, intentional
and undistracting. A boom microphone that can be seen in the top of
frame is disturbing, particularly if it moves to follow dialogue. On
video recordings and film this blemish can often be painted out
electronically in post-production, which also (conceivably) means
that sound coverage might be improved deliberately, with the paint
job planned for.
Most microphones used in vision are either personal, fixed below
the neck or held in the hand, in which case they are moved around by
the performer, or they are static, on table or floor stands.
For a speaker who is not going to move, a fixed microphone in the
picture gives better sound. It will generally be angled to favour the
Speech balance
109
speaker (or split between two): if this chosen angle conflicts with the
designer's choice of `line' within the picture the microphone may be
more noticeable, as it will also be in a simple uncluttered design.
The base, for stability and to cushion shocks, needs to be reasonably
large, so it can sometimes look more obtrusive than the microphone.
If it is recessed into a console, the cavity must be designed with
care: if it is fully boxed in it can modify the local acoustics, so one or
more sides of the recess should be left open. The microphone
capsule itself should stand clear of the surface, so in a shot showing
the table it will be in vision, unless it is hidden behind something on
or in front of the table ± which in turn could be more obtrusive than
the microphone. Alternatively, a specially constructed console might
have its front continue upward a little beyond the level of the
desktop. An acoustically transparent top can be included in the
design.
[7.12] Television desk and
microphone
To avoid bumps, the microphone is
on a floor stand, emerging through a
hole (above), or behind a false front
(below).
A microphone placed to one side may be neater than one immediately in front; it should go on the side to which the speaker is more
likely to turn. However, any turn the other way may go right
off-microphone, so it helps if any neighbouring microphone can
cover this.
Where many people are spread over a large area of studio, a separate
microphone is sometimes used for each individual ± but quality
demands that as few as possible be open at any one time. So
the right ones can be faded up in time, it should be made clear
who is to speak next, either from a script or by introduction by name
or visual indication. When the output of individual microphones is
recorded on multitrack, an improved (or corrected) mix can be left
until later.
The use of hand microphones is a widely accepted convention in
television reporting. It is convenient for the reporter plus cameraman
combination, and also because the microphone can be moved closer
if surroundings become noisier. In extreme conditions lip microphones have been used in vision. But hand-held microphones can
easily be overused, particularly where they make an interviewer look
dominating and aggressive. Alternatives include the use of small
personal microphones (often on a lapel) or, for exterior work, gun
microphones.
In interviews or discussions, participants using personal microphones can be steered by panpots to suitable positions on the left
or right of frame, roughly matching those in the widest shot. In an
unscripted discussion it is better to make no attempt to match
position in frame: this makes editing difficult and in a television
picture a small discrepancy is not likely to be noticed. The pictures
establish a `line' that the camera should not cross, and sound can
follow the same convention. Two speakers (as in an interview) can
be placed roughly one-third and two-thirds across the stereo image.
But note that as two omnidirectional microphones are moved
towards each other, they will `pull' closer still, as each begins to
pick up the other subject, and this may need correction by widening
the image.
110
Speech balance
The main benefit from stereo output here is that separation of voices
from each other and from sound effects and reverberation (which
should be full width), as well as from any noise, aids intelligibility
and adds realism. A separate stereo pair can be directed into the
scene and faded up just enough to establish a stereo background
effect.
[7.13] Video camera with short gun
microphone
This ensures that loud effects in the
field of view are recorded. For
television sports, a full-scale gun
microphone is used to capture
more distant effects.
Using gun microphones
A gun microphone is most useful when filming outdoors, held just
out of frame and pointed directly at each speaker in turn. Camera and
sound must co-operate to avoid getting the microphone in picture: its
big windshield is obtrusive even when still, and more so if it moves.
An advantage of this microphone is that it allows great mobility,
though not of course for the speaker to turn right away from the
microphone. Accidentally off-microphone sound drops to low level
in exterior work, and gives bad quality in interiors. The gun microphone
can, however, be used in the relatively dead surroundings of a television
studio (in an area well away from strongly reflecting surfaces) for
picking up voices at a distance. But in general its directional qualities
are not trustworthy in normal interiors or even reverberant exteriors.
The short gun microphone can also be attached to a video camera,
pointing along its line. It picks up many sound effects adequately
within the field of the average wide-angle lens, but good voice
balance only when the camera is close in. One application is to
cover events or news when no separate recordist is available.
When a highly directional microphone is used to cover effects such
as tap dance, or to follow the action in sports (to some of which it
adds great realism), it is often best to filter frequencies below about
300 Hz. A gun microphone with bass-cut has sometimes been used in
conjunction with other microphones to cover large press conferences,
supplying presence that other distant microphones lack. For this, a
parabolic reflector could also be considered: it would be out of frame,
or possibly hidden behind a gauze that is lit to look solid. Remember
that a reflector concentrates, and so amplifies, the wanted sound.
Another, more specialized microphone for full-frequency pick-up of
distant sources is the adapted-array (digitally processed) combination
microphone: this has a forward lobe 8 across, switchable to 8 30 .
Using personal microphones
For location work, an electret is often fixed to the clothing of each
speaker ± but take care if two are brought close together, as this can
affect the balance. (Older, heavier personal microphones were hung
Speech balance
111
around the neck, and these may still be encountered.) But always
remember that a boom would give better sound ± often much better.
The electret can often be hidden, perhaps behind a tie ± though some
synthetics cause crackles, and a woolly tie can produce an equally
woolly response. Whether it is concealed or not, it is best to ensure
that the means of suspension is also invisible or low-key ± for
example, by using a clip or cord of the same tone or colour as
clothing against which it may otherwise be seen. Some capsules
are available in a range of colours, such as black, white, silver or tan
(flesh tone). Otherwise a slip-on cover of thin, matching material
can help.
The user may have to avoid large or sudden movements, especially
those that disturb the drape of clothes (as sitting down or standing up
does with a jacket). However, normal movement should not be
discouraged unless necessary. A speaker who turns right away
from the camera may still be heard clearly ± though turning the
head to look over the wrong shoulder may cause speech to be offmicrophone.
The cable may be concealed beneath a jacket for a tight mid-shot
(although a personal microphone may not really be needed for so
close a shot). In the more general case it will go to a hidden radio
transmitter or continue on its path beneath the clothing, down a
trouser leg, or inside a skirt ± to a connector not far from the feet,
so the wearer can easily be disconnected during breaks. If the user is
going to walk it may be best to attach any cable (and connector) to a
foot, perhaps by tying it in to a shoelace. But to walk dragging a
length of cable is awkward, and perhaps visibly so. It may also cause
cable noise ± reinforcing the case for a radio link.
Because of its position, the sound on a lapel microphone may be
characteristically poorer than for better angles of pick-up. This
makes it difficult to match, so once started, it is best to persist with
it until there is either a marked change of location (e.g. exterior to
interior) or a long pause before the same voice recurs. To a lesser
degree, this is also true for many other microphones. The effect of
disregarding this is obvious on many television news broadcasts.
A further disadvantage of personal microphones when compared
with booms is that there is no change of aural with visual perspective. This is more important in drama than in other productions.
But they do usually offer reasonable intelligibility in fairly noisy
surroundings, or in acoustics that are brighter than usual for television. The quality gains a little from good reflected sound.
Worn by film or television actors, they will generally be hidden. For
ensemble scenes in some films, all major players have been recorded
separately, giving the director more options in post-production.
For a hand-held microphone in vision, the sound quality problem
goes away, to be replaced by the obvious visual problem, especially
if it partly covers the performer's face. Visually, short-stub UHF
radio microphones are better than the less common trailing-antenna
VHF, and both are usually better than trailing cable. These and
112
Speech balance
microphones on headsets are more often used for music, as described
in Chapter 8.
Radio drama
Although virtually extinct in the United States, some American
readers may find interest in this account of what they may think of
as a lost art. In fact, radio drama is still vigorously alive in countries
such as Britain, where both the skills and the facilities for it still
exist. The techniques are of course also potentially of use for other
complex voice recordings. In some countries (again including
Britain) they are also used for educational radio.
In radio drama the acoustics of a studio are used creatively, the
`scenery' being changed by altering the acoustic furniture around the
microphone. Specialized studio suites are divided into areas with
different amounts of sound-absorbent or reflecting treatment; each
area provides a recognizably different quality of sound. The live
areas offer a realistic simulation of the actual conditions they are
supposed to represent; the dead areas, less so. If similar microphone
types are employed in the different acoustics, the voice quality does
not change but is matched from scene to scene as the action
progresses.
Since one factor in the aural perception of studio size is the timing of
the first major reflection, delay may sometimes be used to increase
apparent size without creating excessive reverberation. But the quality of the reverberation itself also plays an important part. A small
room sounds small because first reflections are very early. A combination of this with reverberation time suggests both size and the
acoustic furniture of a room: a sound with a reverberation time of
1 second may originate from a small room with little furniture
(therefore rather bright acoustically) or from a very large hall with
a rather dead acoustic. The two would be readily distinguishable.
[7.14] Drama studio suite
The main areas in a layout of the kind
used for broadcast drama: some
studios of this kind still exist and are
used for their original purpose. In this
example a virtue has been made of
the irregular shape of a site on which
some of the walls are not parallel,
so there is less standing wave
coloration. 1, Main acting area, with
`live' and `dead' ends of the studio
that can be partially isolated from
each other (and their acoustics
modified) by drawing thick double
curtains across the middle. 2, A `dead'
room, with deep absorbers on all
walls. 3, Control room (for acoustic
isolation there will be double doors).
4, Machine room.
For example, a `bathroom' acoustic would be provided by a small
room with strongly sound-reflecting walls, and can be used for any
other location that might realistically have the same acoustic qualities. Similarly, a part of the main studio (or a larger room) with a live
acoustic can be used to represent a public hall, a courtroom, or a
small concert hall ± and when not needed for drama can also be used
as a small music studio. When going from one type of acoustic
to another, balance (in particular, distance from the microphone)
can be used to help to differentiate between them; for example, the
first lines of a scene set in a large reverberant entrance hall can be
played a little farther from the microphone than the action which
follows ± it would be hard on the ears if the whole scene had heavy
reverberation.
For monophonic drama, ribbon microphones have been widely used:
their frequency range is well adapted to that of the voice, and with
Speech balance
113
working distances of up to 1 m (about 3 ft) give a generous account
of studio acoustics. Actors can work on either live side of the
microphone, standing two or three abreast, and can move `off' either
by moving back or by easing round to the dead side. The conditions
are comfortable, flexible and easily understood. The actor is working
with the acoustics.
Still for mono, for a large, open acoustic with many actors, such as a
courtroom, public meeting or council chamber scene, an alternative
to the ribbon is a cardioid suspended above the action. Working
distances and speed of approach or departure are set by listening to
the acoustic balance. An additional hypercardioid microphone can
be set in for quiet asides that require little reverberation.
[7.15] Simple layout for drama
(mono)
A simple set-up using three ribbon
microphones for dramatic work.
1, Narrator (close to window).
2, `Open microphone' ± for normal
indoor acoustics. 3, `Tent' ± to
represent outdoor quality. This is a
much less versatile layout than can be
obtained in a multiple-acoustic studio,
and the results sound more stylized.
This might be more acceptable for
broadcasting directed to schools,
because: (a) school audiences often
have poor listening conditions; (b) the
number of actors is likely to be small;
and (c) educational broadcasting
generally has a low budget.
For either mono or stereo, if there is a narrator as well as the `livingroom' action, the two must be recognizably different. There will, of
course, be differences in the style of acting, and the use of the fader
at the start and end of each scene will encapsulate it, and so further
distinguish between the two. But acoustics can also emphasize the
differences: if the scene is played well out in an open studio the
narrator can be placed near one of the walls and at a closer working
distance to the narrator's microphone, which may be of a different
type ± often a hypercardioid. The distance from this to the microphone
covering dramatic action must allow for any rapid turn from one to the
other (although more often they will be recorded separately). A screen
near the narration microphone may help to increase the contrast. Note
that even for stereo drama, a narrator will often be mono.
It is sometimes necessary to mix acoustics: to have two voices in
acoustically different parts of the same studio, but taking care to
avoid spill. There is more danger of spill where the voice in the
brighter acoustic is louder. Suitable placing of directional microphones will help, but may not prevent the spill of reverberation from
live acoustic to dead. Thick curtains running out from the walls may
help to trap the sound ± or screens may be used for that ± but in
extreme cases a separate studio (or the studio lobby) or pre-recording could be used.
Screens are used to add realistic touches. For example, the interior of
a car contains a mixture of padded and reflecting surfaces. A `box'
of screens, some bright, some padded, gives a similar quality of
sound.
Open-air acoustics
Open-air acoustics, frequently required in drama, are characterized
in nature by an almost complete lack of reverberation. Even when
there are walls to return some of it, the total reflected sound at the
source is low, and is likely to produce only a small change in quality.
It is, of course, possible to simulate this realistically by dividing off a
section of the studio and padding it deeply to create almost completely dead acoustics, i.e. a dead-room. There are clear advantages:
114
Speech balance
.
.
.
.
.
it provides the best possible contrast to other acoustics in use, so
it makes a wider range of sound quality possible;
the muffling effect of the treatment causes the performer to use
more voice and use the same `edge' on it as in the open air;
effects recorded outdoors blend in easily;
off-microphone voices and spot effects can be blended with
on-microphone voices (in mono by using positions round toward
the dead side of a directional microphone);
the amount of space round the microphone is greater than within
a tent of screens.
Unfortunately, there are also disadvantages:
[7.16] Tent of screens with
bidirectional microphone
In a double-V of screens it is better
that sound should be reflected twice
on each side (above). Standing waves
cannot form unless there are parallel
surfaces directly opposite each other
(below).
[7.17] Use of screens
Double reflection of sound in a V of
screens set at an acute angle. For any
particular path some frequencies are
poorly absorbed, but they are
different for the various possible
paths. Sound that is reflected back to
the region of the microphone has less
coloration with this arrangement than
from a broad V.
.
.
.
In fact, the true dead-room is not a practicable proposition, and a
compromise has to be found. Where a script is set almost entirely out
of doors, it is more convenient to adopt a formalized style with an
acoustic that is hardly less reverberant than a normal speech balance.
However, if there are only a few isolated lines set outdoors, something close to the dead acoustic is more appropriate. In general, try to
set the average acoustic demanded by the script as one that is close
to a normal speech balance, and arrange the others relative to this.
Of the various possible sound-deadening systems, try to avoid those
that give a padded-cell effect. This is, literally, what a thinly padded
dead-room sounds like, and the same thing can all too easily be
repeated with a tight cocoon of screens. For monophonic drama,
keep it simple: try two Vs of screens, one on each side of a bidirectional microphone. Some points to remember with this layout are:
.
.
.
.
[7.18] Speaking `off' a bidirectional
microphone
No performer should go further off
than this, or turn further into the
screens. More `distant' effects must
be achieved by changes of voice
quality or by adjusting the
microphone fader.
such an acoustic can soon feel oppressive and unreal;
the transient sounds in speech tend to overmodulate, even if the
general level is kept low;
completely dead sound is unpleasant to listen to for very long.
Keep the screens as close to the microphone as is conveniently
possible; the actors should accept some restriction on their movements. This keeps the sound path lengths short and damps out
reverberation (and coloration) as quickly as possible.
Avoid setting the screens in parallel pairs: it is better to keep the
angle in the Vs fairly acute. This helps to reduce bass as well as top.
Also, actors should avoid getting too far back into the apex of the V.
Exit and entry speeches can be awkward. In nature, distant speech is
characterized by a higher voice level arriving at a lower final
volume. In the studio, a voice moving off is likely to be characterized
by a sudden increase in studio acoustic. So stay inside the screens at
all times, and avoid directing the voice out into the open studio.
Off-microphone (i.e. `distant') speeches may be spoken from the
screens, speaking across the V. If this is still too loud to mix with
other voices, the actor should change voice quality, giving it less
volume and more edge ± and should not deaden the voice by
turning farther into the angle of the screens: this produces a
cotton-wool muffled quality.
The effect of screens is not that of the open air. But it is not a normal
live acoustic either. It is different, and acceptable as a convention.
Speech balance
115
Stereo drama
[7.19] Stage for stereo speech
1, If crossed hypercardioids are used,
voices in these two positions, roughly
opposite to each other in the studio,
appear to come from the
loudspeakers. 2, Most of the action
takes place in this area, at a distance
of about 2±3 m from the
microphones. 3, More intimate
speech may be closer.
For stereo drama, coincident pairs can be used for each main acting
area. Crossed cardioids or hypercardioids give the best use of space.
Where actors are unfamiliar with the technique, the studio floor may
be marked up with strips of white adhesive tape on the carpet
showing the main limits of the acting area, together with any special
positions or lines of movement. Movement in a straight line on the
sound stage is simulated by movement along an arc of the studio: the
acting area can therefore be defined by an arc with the microphone
at its centre, or by arcs roughly limiting it. Such arcs can be drawn
either as having constant radius (for convenience) or at distances that
give constant A ‡ B pick-up ± which is better for more complex
situations, such as group or crowd scenes in which the relative
distance of different voices is important. From the microphone polar
diagrams it can be seen that for a crossed cardioid this arc is farther
away at the centre line and loops in closer at the sides. In addition,
the studio itself has an effect: unless the acoustic treatment is evenly
balanced, the line may `pull' closer to the microphone on one side
than on the other. Sound stage positions may also be offset laterally
from studio positions: the whole studio may seem to `pull' to the left
or right.
Positions `in the loudspeaker' on the two sides may also be set:
for cardioids crossed at 90 , these would in principle be at the
270 limits of the theoretically available stage. In fact, the cardioid response of practical microphones is imperfect at the rear, so
the best `in speaker' positions would usually be found at about
200±220 separation. It is often convenient to reduce this to, say,
180 , by setting microphones with variable polar response one
step towards hypercardioid, or using a pair with the narrower
response.
In mono, a long approach from a distance can be simulated by a
relatively short movement by the performer, who simply walks
slowly round from the dead to the live side of a directional microphone. In stereo, long approaches do really have to be long, and must
be planned in terms of what space is available: a decision on this in
advance will probably determine studio layout.
In stereo, balances that are more distant than those normally used for
mono may be necessary, in order to avoid exaggerated movements.
For safety, the nearest part of the real stage in which there will be
lateral movement should be at least as broad as the sound stage. In
consequence, the studio must be somewhat deader than for a comparable scene in mono, and reverberation added if brighter-thanaverage sound is required.
For this, use either stereo reverberation or a second stereo pair
within the same studio. If a coincident pair is chosen for this, it
could be about 1.2 m (4 ft) above the first pair. This provides some
degree of automatic compensation as a speaker moves in close (but
note that a tall person will be closer to the upper microphone than
116
Speech balance
someone shorter would be). An alternative arrangement is to use a
spaced pair pointing back to the end of the studio that is away from
the stage.
Whereas a small studio for drama in mono may have several acting
areas, the same studio used for stereo may have room for just a
single stage (this allows for a reduction in reverberation and an
occasional need for deep perspective). Therefore there may be good
reason to record all scenes that have one basic acoustic before
resetting the studio for the next, eventually editing the programme
to the correct running order. If very different acoustics are required
in the same scene ± e.g. a distant open-air crowd scene including
studio `open-air' voices, mixed with close `interior' conversation ± a
pre-recording has to be made, the studio reset, and the second part
layered in to a playback of the first.
To simulate stereo exterior scenes in a studio, a variation on the
same basic studio acoustic may be unavoidable. Cutting bass on the
microphone output or at the mixer may help, but the result of that
depends on the acoustic coloration of the studio in use. For the effect
of a more distant voice in the open air, the actor may help by staying
well forward and projecting and thinning tone of voice, or, alternatively, by turning away from the microphone and speaking into a
thick screen. Ideally it is bass that needs to be absorbed more.
Stereo drama problems
Dramatic conventions for stereo are well established. One is that
movement should occur only during speech from the same source or
while other sounds (e.g. footsteps) are produced; certainly, moves
should never be made in a short pause between two speeches, except
for blind man's buff comic or dramatic effects ± easily overused.
Problems that are met in mono drama may be greater in stereo. For
example, extraneous noises are worse because:
.
.
.
more of the studio is open to one microphone or the other;
with a more distant balance of voices, the signal at the microphone is at a lower level; channel gains are higher and therefore
noise is amplified more;
noises may be spatially separated from other action, and therefore
less easily masked.
Special difficulties are raised by a need for artistic compatibility
between stereo and mono. Consider, for example, a conversation
between two people trying to escape unnoticed through a crowd:
they talk unconcernedly as they move gradually across the stage
until eventually they are free of the other actors and can exit at the
side. The mono listener to the stereo drama needs as many positional
dialogue pointers as would be given if the production were not in
Speech balance
117
[7.20] Stereo drama studio layout
In many studios (as here) there may
be asymmetry in the acoustic
treatment, distorting the sound
stage. In this example, floor markings
show minimum working distance
around the microphone pair. This is a
layout that was designed by its
director (and confirmed by aural
judgement) to emphasize extreme
side positions and allow long
approaches from left (1) and right (2).
3, Drapes screening live studio area
beyond. 4, Announcer or narrator,
centred.
stereo. Indeed, in mono the microphone would probably stay with
the main subjects and other near voices would drift in and out of a
clear hearing distance; in stereo a convention like this would be
more difficult to establish. Imagine or listen to both stereo and mono
versions of the chosen convention and decide whether changes could
or should be made for the benefit of either the stereo or the mono
audience ± or whether to disregard potential mono listeners (who
may be an important part of a radio audience).
A more extreme case might be a telephone conversation: a theoretical possibility would be to split the stage, putting one actor in
to each loudspeaker. This might work well in stereo, but for mono
would be meaningless. In this case artistic compatibility requires
that the voice of one of the actors must be filtered. The filtered
voice could still be put in one of the loudspeakers, but the other,
`near' voice would probably be closer to the centre of the sound
stage and in a normal stereo acoustic ambience, but offset to
balance the other.
Audience reaction
Audience reaction is a subject that links this chapter with the next. It
is a vital element of many comedy presentations; for music broadcasts it provides a sense of occasion that a home audience can share.
The two differ in that for the first, the performers' and the audience
microphones must be live at the same time, while for music performances they are sequential. For all types of programme, applause
(and other forms of audience reaction) must be clear, without being
unbalanced ± that is, individual pairs of clapping hands should not
be loud or assertive.
To achieve this, several possibly conflicting requirements must be
met. The audience must be reasonably large ± ideally, at least
several hundred strong (although a smaller audience can in principle
be doubled by using a short delay). The microphone coverage must
118
Speech balance
not be so distant as to lose clarity. In addition, there must be
adequate separation between the audience and other live microphones. Two or more audience microphones will often be used,
and for musical performances these may also act as `space' microphones. They are pre-mixed on the control console and fed to a
single group fader.
It may help to have a compressor or limiter before the main audience
fader. A further limiter or compressor can be inserted after mixing
audience and dialogue, and used in such a way that speech pushes
down the audience reaction as the voice itself is compressed.
In stereo, applause is best spread across the full sound stage. It must
be picked up on the main and announcer's microphones only in such
a way that controlling its level does not pull the applause to one side.
[7.21] Audience loudspeaker and
microphone
1, Line loudspeaker. Except at low
frequencies the elongated source
radiates at right angles to its length,
and low frequencies are reduced by
bass roll-off. 2, The audience
microphone is angled with its dead
side towards the loudspeaker on the
bar, and the line loudspeaker does not
radiate sound towards the
microphone. Attached to the same
bar, their relative positions are fixed.
[7.22] Avoiding spill from public
address loudspeakers
Even when a microphone is nominally
dead at the rear it is usually safer to
set audience loudspeakers at a rear
angle of 45 , where the frequency
response is cleaner.
The audience reaction to a comedy show will obviously be poor if
the words are not clearly audible. This requires PA loudspeakers
above the heads of the audience, fed by one of the auxiliary send
channels available on studio consoles. A line-source loudspeaker
radiates most strongly in a plane at right angles to its length, and a
microphone placed outside this plane will receive its middle, and
high-frequency sound only by reflection. Unless they are well away
from these loudspeakers, the audience microphones must be directional and dead-side on to them. Use of the rear axis of a unidirectional microphone is generally avoided, because in many cases the
frequency response is most erratic in that direction. Instead, it is
safer to place the microphone so that the nearest loudspeaker is at
45 to its rear axis. The PA send is in this case a specially tailored
mix of only those channels required for intelligibility. Bass can be
cut or rolled off, and equalization introduced to increase clarity and
avoid coloration.
In television, for which the performers' microphones may be set
well back from the action they are covering, the PA loudspeakers
must be brought as close over the audience as is practicable. The
loudspeaker volume can be raised by 3±4 dB without increasing the
danger of howlround if the frequency of the public address sound is
shifted (usually downward) by a few hertz, using a pitch changer.
If the response from some particular individual in the audience
becomes obtrusive, fade down the microphones gently, in turn, until
that nearest to the problem is located, then reset it accordingly.
The volume of applause at the end of a performance must be held
back a little: prolonged clapping at high level is irritating to the
listener, who may be impelled to turn it down. On the other hand,
laughter within a programme may have to be lifted a little: it could
be held down between laughs in order to avoid picking up coloration, raised quickly as the laugh begins, and then eased back under
subsequent dialogue just enough to avoid significant loss of clarity
in speech. The performers usually adapt rapidly to audience
response, and sensitive control should anticipate the progress of this
interaction.
Chapter 8
Music balance
There are two main types of music balance. One is `natural' ± that is,
music heard in natural acoustics ± and is generally preferred for
classical music. For a single instrument or for a group having perfect
internal balance, one microphone (or stereo pair) may be sufficient.
But such perfection is rare: a singer may be overwhelmed by all but
the quietest of pianos; an orchestra may be dominated by its brass.
So even `natural' balance often requires additional microphones.
Most music is balanced in stereo, for which the techniques may be
further subdivided into those using coincident pairs and others with
spaced microphone arrangements. Mono is still used in parts of the
world where, for economic or language reasons, stereo broadcasting
is restricted. (In television, if several language versions are required,
this may take up audio channels that could otherwise be allocated
to stereo.) But mono is also used for individual instruments or
sections within a stereo balance. Further treatment of individual
components of the whole sound is normally used only for purposes
of correction.
At the other extreme, natural acoustics are all but eliminated, to be
replaced by stereo artificial reverberation. This is a multimicrophone
balance that at one time was used for nearly all popular music. It
employs many close microphones, predominantly mono, though
with some stereo pairs. Each component has its own `treatment'
and its metered dose of artificial reverberation (AR) before being
steered to some appropriate place within a composite stereo image.
This whole process may be extended over time by multitrack recording before the final mix-down.
Here, too, there are less extreme versions ± in part, with the rediscovery of the virtues of natural acoustics for some components of the
balance. In fact, many balances are somewhere in the range between
the two extremes. Each of the main types of balance will be considered, first in general terms, and then by looking at balances for
120
Music balance
individual instruments or sections that could be used in either of
them.
Music in 5.1 surround sound is generally balanced initially as for
stereo, but with reverberation distributed all round, although some
mixes may also distribute some of the sources. Sources such as a full
orchestra, which may warrant the use of microphones to pick up
reverberation, can be balanced for multichannel recording designed
directly for 5.1.
`Natural' balance
Here the most important component of balance is that between direct
sound and natural reverberation. In a natural balance taken on a
single microphone (or coincident stereo pair), its distance from the
source depends on both the polar response of the chosen microphone
and the acoustics of the studio. A flexible method employs a condenser microphone (or pair) that has a remotely switchable polar
response. With this, try various distances for each polar response and
listen how clarity (of individual components and of melodic line)
compares with blending, or warmth of tone. If possible, set several
microphones (or pairs) of the same type at different distances, in
order to make direct comparison tests. This is really the only practicable way of determining whether the microphone is within the
distance at which reverberation begins to swamp the direct sound
and early reflections.
A close balance sounds more dramatic, but may be difficult to live
with. Where the acoustics are difficult, it may be necessary to accept
[8.1] Frequency ranges
A microphone with a frequency
response that matches that of the
sound source will not pick up noise
and spill from other instruments
unnecessarily. The lowest frequency
for most instruments is clearly
defined, but the highest significant
overtones are not. At high
frequencies, an even response is
usually considered more valuable
than a matched one.
Music balance
121
a brilliant close balance, with artificial reverberation if desired: this
can always be added, whereas a bad acoustic cannot be subtracted.
The spread of orchestra or ensemble must be matched to polar
response and distance. With a distant balance a bidirectional response may be possible, but as the microphone is moved in, a
broader pick-up is required. In many cases the best mono is achieved
by a cardioid with a clean, even response (in which the polar
response is independent of frequency). Omnidirectional microphones will have to be nearer still ± perhaps noticeably closer to
some instruments than to others. They are therefore more suited to
spaced-microphone techniques.
As spotting (accent) microphones, smooth cardioids or hypercardioids
are again favoured. Bidirectional ribbons, too, can offer an excellent,
smooth response in all but the extreme top ± which in any case lies
beyond the range of the human voice and many instruments.
When they have high-quality loudspeakers, many listeners prefer a
more distant and so more blended sound than some balancers enjoy.
But for listening on poorer equipment, greater clarity compensates
for some of the deficiencies: at least it is possible to hear the main
elements of the music. For this reason (and also because it helps to
motivate close-ups), a drier balance is often used on television.
The arrival of an audience can change a balance dramatically.
Experience will show how much too live the balance must be in
rehearsal for it to be correct with a full hall. It will usually be safer to
have some separate means of controlling this, either by combinations of microphones or by adding artificial reverberation.
Music studio problems
A `natural' balance records the characteristics of the studio just as
much as those of the musicians, rather as though the studio were an
extra member of the orchestra. With stringed instruments, very little
of the sound we hear is directly from the strings; almost all is from a
sounding board to which the strings are coupled, so the character of
each instrument depends on the shape and size of that radiator. In the
same way, the studio acts as a sounding board to the entire ensemble: its shape and size give character to the music. But there is one
major difference. By uniformity of design, the character of all
instruments of any particular type is roughly the same. But music
rooms, studios and halls differ widely in character: no two are
the same.
In the natural balance directional microphones are angled to pick up
not only direct sound, but also the early reflections and reverberation
from all three dimensions. In a large auditorium, the volume of
direct sound will diminish with distance from the source, while
reflected sound is at much the same volume throughout the entire
space ± so there is a critical distance beyond which reverberation
122
Music balance
dominates the sound heard. Plainly, the best place for a microphone
(or microphones) will be inside that critical distance, so that the
balance is enriched but not overwhelmed by reverberation. In an
excessively live interior (a cathedral, perhaps), an audience seated
well beyond the critical distance cannot hear a satisfactory acoustic
balance for most music, let alone speech, and this is certainly no
place for a microphone.
At the other extreme, in a studio that is unusually dead for its size, a
more open directional microphone response may help ± a broad
cardioid, perhaps. There can even be times when even a distant
balance with an omnidirectional microphone does not do enough.
Some percussive instruments, such as timpani, continue to radiate
sound with a decay characteristic a bit like that of reverberation. If
they stand out in too dead a hall, artificial reverberation should
be added.
A typical television or film studio is a dead studio on the grand scale.
The same problems may arise in microcosm when one or two singers
or instruments are recorded on location, say in a domestic environment that is not designed for music making. In this case, taking a
microphone to the other side of a small room with heavily damped
acoustics will only emphasize their inadequacy. The only effective
solution may be furniture removal: if possible get rid of heavily
padded furniture such as armchairs. Roll up rugs and draw aside soft
drapes. And avoid having too many people in the room.
Theory suggests that the central line should be avoided, because in
the middle of a room the even harmonics of the eigentones ± the
basic resonances of the room ± are missing, thereby thinning out its
acoustic response. Fortunately, in practice, few studios seem to
exhibit serious eigentone problems in music balance. In fact, stereo
microphones are often placed on the centre line of concert halls and
rarely with ill effect (but note that concert-hall eigentones are very
low pitched).
For monophonic balance (which may be a component of stereo) in
difficult acoustics, a directional microphone gives the greatest control as it allows an extra dimension for experiment: the angle can be
varied, as can the distance from the source and position within the
room, and on some microphones the polar response too.
Stereo music balances that are based primarily on coincident pairs of
microphones set initially with cardioid elements angled at about
130 may also be modified either by changing the polar diagram,
or by changing the relative angle of the elements, or both.
In particular, there are two reasons why the double figure-of-eight
system is rarely used. The most important is poor compatibility with
mono: there are likely to be some listeners to a stereo broadcast or
recording who hear it in mono, and if the A B signal is large they
lose a significant part of the information. In fact they lose much of
the reverberation, which is strongly represented in the A B component of a double figure-of-eight. Another problem is that reverberation is picked up on both back and front of the microphones in
roughly equal proportions, so that at low frequencies there is a
Music balance
123
phase-cancellation effect that takes some of the body out of the
stereo reverberation, making it thin and harsh. This is noticeable in
a concert hall and marked in the more reverberant surroundings of,
say, a cathedral. Switching the polar diagram even a quarter of the
way towards cardioid reduces this significantly.
One microphone or many?
The extreme alternative to studio acoustics and the natural blending
of sounds is the multimicrophone balance in which an essential, but
not necessarily conventionally natural, quality is sought from each
instrument or group and blended in the mixer. Neither technique is
any more `right' than the other, but opinion has tended to polarize to
the view that most classical music is best balanced on fewer microphones, and that popular music demands many.
There is a rightness about this: the `simpler' microphone technique for
classical music reflects the internal balance of the orchestra and existing acoustics that the composer wrote for, whereas the top 20 hit of the
moment is more likely to reflect the search for novelty and opportunities for experiment. But the question `one microphone or many?' is
still open for many types of music that lie between the extremes.
Before choosing a `natural' balance, consider how the sound source
is distributed. Even if the acoustics are suitable, can it be arranged
that the sound source lies within the pick-up field of a single microphone position? If it can, then given a satisfactory instrumental
layout, careful microphone placing can ensure that the distance
between the microphone and each source is exactly right.
Individual instruments and
groups
Orchestral players and pop musicians play instruments that have
much more in common than in difference. All orchestral instruments
are fair game for popular music, but there are others that rarely make
the journey in the opposite direction, notably electric guitars and
synthesizers, but also some older sounds such as those of the piano
accordion or banjo. But it is clear that for all types of music balance
we are dealing with the same basic kinds of sound sources, which
can be treated in a range of different ways, but for which the various
possible treatments required in different types of music overlap.
A natural orchestral balance may require reinforcement of a particular instrument or section, or in a stereo balance mono spotting may
be needed to make a position clear: both of these demand close
microphone techniques that approach those used in multimicrophone balance.
124
Music balance
A question that is still debated is whether it may sometimes be
desirable to delay the signal from a spotting microphone to match
its apparent distance to the actual distance from the main microphones. To estimate the time involved, roughly measure the distance
from the main to the spotting microphone in front of the instruments
it covers: as a rule of thumb, this distance in feet (or one-third of the
distance in metres) gives the number of milliseconds of delay to go
in the spotting channel. A more accurate calculation is: at 20 C,
delay in seconds equals number of metres divided by 344. As an
example, if a main pair is 10 m further back than the spotting
microphones, the delay would be 29 ms. A tiny delay (of, say,
10 ms) would very likely be masked in the mix by sound from the
main pair, but nearly 30 ms is too much. This suggests that in a
distant orchestral balance, a delay should be inserted, while for most
other (closer) balances it may be disregarded.
For what counts as a `close' balance, there may be differences in
terminology. In classical music a distance of a metre (several feet) is
`close' or `very close', depending on the instrument; for pop music,
closeness is measured in centimetres (inches) or replaced by contact.
But for all of these, there is no difference in the basic problems and
possibilities, and in the following sections they are explored first for
individual instruments, sections and combinations, and then for full
orchestras and bands.
Most instruments belong to one of three main groups:
.
.
.
Strings ± plucked, bowed or struck, and coupled to a sounding
board acting as a resonator (or from which vibrations are derived
electrically and fed to a loudspeaker).
Wind ± where the resonator is an air column excited by edge tone,
vibrating reed, etc.
Percussion ± where a resonator (often with an inharmonic pattern
of overtones) is struck.
Note that while there are general principles that may apply to each of
these, some players and some instruments produce different sounds
from others of the same apparent style or type (think, for example, of
double basses and their players). Note also that any balance that
actually works does so only for the particular microphone set-up
used, and in all but close balances, only for a particular studio. It
may even be affected by the weather: if it is very dry, a good natural
balance is more difficult, because high frequencies are absorbed by
dry air.
Violin, viola
[8.2] Radiation pattern and
microphone position for violin
1, Low-frequency radiation. 2, High
frequencies. 3, Move along this arc
for more or less high frequencies.
4, Move closer for clarity, away for
greater blending.
A violin radiates its strongest concentration of upper harmonics in a
directional pattern, so that for all but the lower frequencies much of
its sound goes into a lobe over the heads of the audience. With that
also goes a great deal of the harsh quality, squeak and scrape that
even good players produce. The audience normally hears little of the
Music balance
125
high-frequency sound, as even in the reverberation that may be
heavily attenuated in the air. At its lowest frequencies the radiation
is omnidirectional, but the violin is an inefficient generator of sound
at low frequency owing to its size: the lowest notes are heard mainly
in their harmonics.
For a concert balance the microphone is placed up and well back
from the instrument to add studio reverberation, and frequency
content is controlled by movement in an arc around the player.
The upper harmonics are heard at their strongest a little off-axis,
towards the E string (the top string on the violin). A ribbon microphone can give a good response (though possibly requiring some
reduction in top).
For a balance with microphones for each individual instrument or
group, separation is more important. In this, the violin microphone
will have to be much closer, perhaps at only 60±90 cm (2±3 ft), and
angled directly into it. This balance may be harsh, but treatment will
be available (and expected) on each channel anyway.
[8.3] Close violin balance in pop
music
The player has encircled the
microphone stand with the bowing
arm. A directional microphone is not
subject to much bass tip-up
(proximity effect), as the lowest
frequency present is a weak 196 Hz.
A filter at 220 Hz may be used to
discriminate against noise and rumble
from other sources. Contact
microphones are sometimes used.
The lowest string of the violin is G (196 Hz), but the fundamental is
weak, so even with close directional microphones bass tip-up is
hardly noticeable; indeed, a bass filter (high-pass filter), rolling off
below 220 Hz, may be used to discriminate against low-frequency
sound from other sources. Rather than roll off top, it may be better to
lift a mid-frequency band that makes the violin stand out among
other instruments: try bands around 300 Hz or 1±1.3 kHz and listen
to see what works. The tone quality of the violin will also affect this.
To make a very close violin sound pleasant requires more than the
art of the balancer can supply: it demands a modified style of playing
in which the scratchier components (normally diminished by distance) are much reduced. When using a stand microphone, very
close working may be limited by the athleticism of the player, so a
little under 1 m (3 ft) may be the minimum, especially for pizzicato
passages in which the bow might hit the microphone. But to get
inside that, some players are capable of encircling the microphone
stand with the right (bowing) arm. In this rock or jazz balance, the
higher overtones (`harmonics' seems too gentle a word) will cut
through.
A contact microphone may also be used, but expect the sound to be
different again. If it is attached near the bridge, its output may lack
high frequencies: experiment with the position and judge whether or
not this is an advantage. Alternatively, try an electret clipped to the
player's lapel: this position discriminates against the stridency.
[8.4] Close balance for three desks
of violins
As an alternative to this, for clean
separation it is possible to work with
a separate microphone for each
desk, but that produces a very strong
high-frequency response which
must be discriminated against or
corrected for.
For a close balance on an orchestral violin section, the players sit in
their normal arrangement of two at a desk with the desks one behind
another, and the microphone is slung directly over and slightly
behind the first two players. This should favour the leader enough
for solos, but should not lose the rear players as a close balance
farther forward would. If this balance is still too distant for good
separation in, say, a show band in live acoustics, put a microphone
over each desk and, as usual, apply high-frequency roll-off to counter the harsher quality.
126
Music balance
Even if used as part of a stereo balance, a mono microphone must be
used for close working, as the slightest movement near a coincident
pair becomes gross. For stereo, the instrument or group is steered to
its appropriate position on the sound stage (for a soloist on his own,
in the centre) and full width reverberation is added as required.
For a balance in which stereo microphones are used, they must be
more distant. Slight movements of body and instruments then produce
an effect that is pleasant, unless they noticeably alter the proportions
of sound reaching the microphones. This can happen easily with
spaced microphones, but may still do so with a coincident pair: a
change in reflected sound can produce spurious apparent movement.
Again, reverberation must be of good quality and well distributed.
Violas may be treated in the same way as violins, except that very
close working on directional microphones may produce bass tip-up,
for which a filter at 220 Hz cannot be used, as this is inside the
frequency range of the instrument.
Cello, bass
Cellos and basses differ from violins and violas in that they have a
much larger area of resonator to drive the air, so they radiate far
more efficiently in the lower register. The high, extended spectrum
of resonance that makes the violin shrill and `edgy' if heard from
directly above is scaled down on the cello to the upper middle
frequencies. To get the full rich sonority of this instrument's upper
harmonics, place the microphone more directly in line with its main
radiating lobe. Again, this can be used as an aid to balance. If the
objective is just to add depth to a fuller orchestral sound, a balance
that is near-sideways-on to the instrument may do enough, but for
richness turn the instrument to the microphone, as a cello is turned to
the audience in a concerto.
If, in a raked orchestra, the cellos or basses are mounted on light
rostra (risers), these may act as sounding boards that add unwanted
coloration. Rostra should therefore be solidly braced or acoustically
damped. Strategically placed, the mass of the players also helps. If
microphone stands are used, avoid placing them on the rostra or pay
special attention to their acoustic insulation. In an orchestra, the
basses may be helped by a spotting microphone, placed about 1 m
(3 ft) from the front desk and angled towards the bridges.
For a cello solo the player faces forward, with the instrument a little
off-axis to the player's right. The brightest (and richest) balance will
also be on this line, and shifting off it will reduce the upper harmonics.
A distance of 1 m (3 ft) is likely to be good, but in a multimicrophone
balance may be reduced to as little as 15 cm (6 in), with the usual
compensation for close working if a directional microphone is used.
For popular music the bass, or double bass ± plucked rather than
bowed ± is one of the three basic instruments of the rhythm group;
the others are piano and drums. Of the three, the bass has the lowest
Music balance
127
volume, so is the most difficult to separate. This suggests a close
balance, directly on to the body of the instrument, but specialists are
disinclined to recommend any single balance that will always work,
pointing out that there are great differences between one bass or bass
player and another. The following alternatives have been suggested:
.
.
.
.
.
a cardioid, hypercardioid or figure-of-eight microphone 30 cm
(1 ft) from the bridge, angled down to the strings;
a similar microphone directed towards the f-hole at the side of the
upper strings;
a miniature (`personal') electret pointing up to face the bridge;
a miniature electret wrapped in a layer of foam plastic, suspended
by its cable inside the upper f-hole ± this gives virtually total
separation;
so too does a contact microphone somewhere near the bridge.
The first or second of these balances will require the usual bass rolloff to compensate for proximity effects. In general, the first has the
most to recommend it, because it picks up a percussive attack quality
from the string as it is plucked, and offers fine control over the ratio
of this to the resonance of the body of the instrument. In contrast, the
fourth balance gives a heavy, hanging quality that must be controlled
to a lower level than the others. In stereo pop balances the bass is
always taken mono and steered to its appropriate position.
More strings: acoustic balances
Other stringed instruments, such as classical guitar, banjo, the violin's
predecessors (the viol family) and so on, all radiate their sound in
ways similar to the instruments already described, and the control that
the balancer has over the ratio of top to bass is much the same. Groups
based on stringed instruments are also balanced in a similar manner.
[8.5] String quartet
1, Viola. 2, Cello. 3 and 4, Violins.
Similar arrangements may be made
for wind and larger chamber groups.
This mono balance employs a
bidirectional ribbon (with a frequency
response that offers a good match to
the instruments of this group); stereo
balances might use coincident pairs
of bidirectional, hypercardioid or
cardioid microphones, moving closer
with each successive pair.
For a string quartet, if the players sit in two rows of two, the
microphone can be placed somewhere along a line extending forward and upward at about 30 from the middle of the group. Check
that the distance, together with the angle and position within the
studio, take full advantage of the acoustics. Listen for each instrument in turn as though it were a soloist, and make sure that the cello
can `see' the microphone. Try to combine the clarity and brilliance
of individual instruments with a resonant well-blended sonority for
the whole group. For radio or recording, the scale of width should be
about one-third of the sound stage; for television it would be wider.
As usual, reverberation spreads over the full available width.
As more instruments are added, a similar `natural' arrangement may
be retained, moving the microphone back if necessary in order to keep
all the instruments within its field. In this way we progress through the
various types of chamber group or wind band to the small orchestra.
For the acoustic guitar as part of a multimicrophone balance, for
high output try a position about 15±30 cm (6±12 in) above the sound
hole and somewhat off-axis. Use a condenser microphone for its
128
Music balance
extended frequency and good transient response. Because air inside
the instrument resonates at 80±100 Hz, a microphone near the sound
hole will require 5±10 dB roll-off below 100 Hz. If separation problems persist, use an isolation booth or record later to playback.
Electric guitar family
[8.6] Electric guitar direct
Above: Direct to console. 1, Guitar
cable. 2, DI box. 3, Microphone cable
(XLR). 4, Console. 5, Headphones.
Below: With direct and `amp'
(6) feeds to console.
The electric guitar has its own contact microphone near the bridge,
from which the output is likely to be an unbalanced, high-level
signal, suitable for sending to a nearby `amp' (amplifier plus loudspeaker cabinet). Some mixers also accept this feed, but in any case
the signal can be split by a DI (direct injection) box to send a lowlevel, electrically isolated and balanced signal like that from other
microphones by an ordinary microphone cable to the console. Where
effects are added by the performer, their output is already included
in this feed. If it is felt that the guitar's built-in pick-up does not
have an adequate high-frequency response, an alternative is to add
another contact microphone, again near the bridge. The player
wears headphones.
A direct feed offers the cleanest guitar sound, and avoids any
problems with separation. But if the `amp' is considered to be an
essential part of the musical quality, then in addition (or instead) a
microphone may be placed in front of the loudspeaker, about
10±20 cm (4±8 in) from the cone. `Amps' vary in quality, and may not
have an extended high-frequency response: plainly, this is unnecessary if the pick-up is limited, though there may be a compensatory
peak below a 5±6 kHz tail-off. In line with the axis of the cone, high
frequencies may be too harsh, so try moving the microphone sideways for the best position. If necessary, separate the `amp' from
other instruments with low screens behind the (directional) microphone. If the loudspeaker system is overloaded, reduce the acoustic
volume (or the bass), restoring it in the mix. A bass guitar
may benefit from lift somewhere between 100 and 400 Hz or in
the 1.5±2 kHz overtone range.
In principle, a hypercardioid microphone response is appropriate. If
the volume of sound results in distortion, check whether there is a
switch on the microphone itself to reduce signal level. Some balancers prefer a `rugged' near-omnidirectional microphone that
becomes more directional at high frequencies. This can handle heavy
volumes and, if it has a large diaphragm, this may be used to control
high frequencies.
[8.7] Guitar balance
For popular music, one possible
balance uses both a contact
microphone at the bridge and an
acoustic microphone clipped over
the air-hole.
In some cases the guitar serves a double purpose as an acoustic and
electric instrument. In the simplest balance the loudspeaker is placed
on a box or chair opposite the player, with the microphone between
loudspeaker and instrument. The player adjusts the overall amplifier
level, comparing it with the direct, acoustic sound, and has control of
the loudspeaker volume from moment to moment by means of a foot
control. A bidirectional microphone can be used for this balance,
Music balance
129
with any bass equalization that may be necessary at the distances
involved.
Other balances include taking a feed from the contact microphone
and combining it with the high-frequency response from an acoustic
microphone clipped to the edge of the sound hole, hanging inside it
directly below the strings and angled to discriminate against noise
from the fingerboard. These may be mixed with the loudspeaker
balance, taking different frequency ranges, equalization and reverberation from each. Double-check the combination by monitoring it
on a small loudspeaker (via AFL, i.e. after the channel fader).
Where a feed of any electrical signal is taken to the console (if not
through a DI box then always through an isolating transformer, for
safety), a radio-frequency rejection filter may be needed to eliminate
unwanted police and taxi messages and, in TV studios, radio talkback.
The combination of contact microphone near the bridge, mixed with
acoustic microphone clipped to the rim of the air hole, may also be
used for other instruments that have this layout, such as dulcimer or
zither. On a banjo, try the contact microphone at the foot of the
strings, and clip an acoustic microphone to the bridge for the strings.
Grand piano
[8.8] Piano balance
For comparison of positions, these
show mono balances with
a bidirectional microphone. The
same principles apply to (slightly closer)
hypercardioids and, for some of the
positions, coincident pairs. The best
balance for a grand piano is usually
somewhere along the arc from the
top strings to the tail. A close balance
gives greater clarity; a distant balance
increases blending. Of the positions
shown: 1, Often gives a good
balance. 2, Discriminates against the
powerful bass of some concert
pianos. 3, Picks up strong crisp bass.
A mix of 2 and 3 can be very effective
as a spaced pair. 4, With the lid off,
close and pointing down toward
upper strings, for mixing into
multimicrophone balances.
5, Discriminates against piano for
pianist/singer. 6, Angled down
towards pianist, same effect as
5. 7, One of a variety of other positions
that are also possible in a stereo pair:
experiment rather than rule of thumb
indicates best balance. 8, Concert
balance, `seeing the strings'. 9, By
reflection from lid. 10, By reflection
from the floor: microphones set for
other instruments may inadvertently
pick up the piano in this way.
The piano generates sound in a manner similar to the smaller
stringed instruments: the string vibration is linked to the soundboard.
This radiates almost all sound that is wanted, with just a small
contribution from the strings at the higher frequencies. If the microphone were to be placed beneath the piano this high-frequency
component would be lost.
The pattern of radiation from the soundboard allows some control of
treble and bass: in particular, the bass is at its strongest if the
microphone is placed at right angles to the length of the piano. In
an arc extending from top to tail of the piano, the most powerful bass
is heard at the end near the top of the keyboard. As the microphone
130
Music balance
is moved round, the bass is progressively reduced until it is at a
minimum near the tail. For very powerful concert pianos the tail
position may be best, but with the slight disadvantage of some loss
of definition in the bass, together with reduced volume. A point near
the middle of the arc should be satisfactory, and is a good starting
point for a balance test.
The piano is the first individual instrument that we have considered
that may benefit by stereo spread of the direct sound. But even for a
solo piano the spread should not be too wide: half the stereo stage is
enough, with reverberation occupying the full width.
Microphone distance is governed largely by the desired ratio of
direct to ambient sounds. A fast piece may need clarity of detail; a
gentler romantic style benefits from greater blending. A closer
balance is also required for brilliant or percussive effects, and to
some extent these qualities may be varied independently of reverberation by changes of microphone field patterns, as well as by
distance. A piano used for percussive effect in an orchestra might
need a spotting microphone. Pianos themselves vary considerably:
some are `harder' (more brilliant) than others, so that for a similar
result greater distance is required. Conversely, a muddy tone quality
may be improved by bringing the microphone closer. A new piano
takes time to mature and is better reserved as a practice instrument
until the tone settles down.
In a dead acoustic such as that of a television studio, try using two
microphones: one is placed close to control brilliance, the other
more distant and with added AR, perhaps with a concert-hall duration of delay.
The height of the microphone should allow it to `see' the strings
(or rather, the greater part of the soundboard), so the farther away it
is, the higher it must be. If this is inconvenient, try a balance that
uses reflections from the lid: this is the balance an audience hears at
a concert. For the deepest notes, the piano's pattern of radiation
tends to be omnidirectional; but for middle and top, and particularly
for the higher harmonics, reflection from the lid ensures clarity of
sound.
Depending on the acoustics, cardioid or bidirectional microphones
may be used. A bidirectional ribbon can give excellent results: its
frequency range is well matched to that of the instrument.
As we move closer to an open piano, the transients associated with
the strike tone become clearer; at their strongest and closest they
may be hard to control without reducing the overall level or risking
distortion on peaks. Action noise ± the tiny click and thud as keys are
lifted and fall back ± may be audible, and in balance tests this and
the noises from pedal action, etc., should be checked. On a close
balance for pop music the lid may be removed, unless this makes
separation more difficult.
[8.9] Close piano balance
Directional microphone suspended
over the top strings ± in this case,
not too close to the action.
A bright and still fairly natural balance suitable for the piano in
a rhythm group may be as close as 15 cm (6 in) over the top strings.
As one possibility, a bidirectional ribbon can still be used: its bass
Music balance
131
tip-up at this distance helps rebalance the sound. The best position
depends on the melodic content of the music: one criterion is that the
notes should all sound in the same perspective, which can also be
affected by the manner of playing. An old trick is to add a microphone baffle, perhaps a piece of cardboard stuck to the back of a
ribbon casing, to lift and `harden' higher frequencies, especially of
transients.
[8.10] Piano holes
The focusing effect of holes in the
piano frame offer a mellow quality.
A microphone 50±75 mm (2 or 3 in)
over hole 2 may give a good mono
balance, but those on either side may
also work; so can pairs mixing the
sound at hole 1 with 4, 5 or 6.
Working at such distances, it may be better to use two microphones,
one near the hammers in the middle top, with the other over the bass
strings, and kept separate to give appropriate breadth to the piano
in the stereo picture. For a hard, percussive effect a microphone
(or two, or a coincident pair) can be slung 10±20 cm (4±8 in) over the
hammers. If the lid has to be kept low (on a short stick) to improve
separation, the separate treatment of equalization and AR makes it
easier to cope with the resulting less-than-ideal conditions.
Another possibility uses the focusing effect of the holes in the iron
frame of some pianos: the full-bodied sound over these can easily be
heard by leaning in to them. A microphone 5±8 cm (2±3 in) above
the second hole from the top may give the best overall balance. With
two microphones, try the top hole and the fourth, fifth or sixth hole.
Overload will cause distortion (particularly of percussive transients),
so must be avoided, perhaps by attenuation at the microphone.
For a fully controlled mono contribution to a group, wrap a small
electret in foam, place it on the frame at the farthest point and close
the lid. Or, for a clear pick-up with good separation, use the lid as the
hard surface for a boundary microphone: to retain some bass, tape it
on the open lid near the middle of the curve. For greater separation
from noisy neighbours, put flats (gobos) between them, or drape a
blanket over the lid to cover the opening. If all else fails, move the
piano to an isolation room (if there is one) or, for a more natural
piano balance, overdub it later.
[8.11] Piano and soloist on one
microphone
Balance for the piano first, and then
balance the soloist to the same
microphone. The piano lid may have
to be set in its lower position.
For stereo, use a coincident pair.
[8.12] Piano and soloist side by side
Balance using two ribbons (1) or
stereo pair (2) and hypercardioid (3).
Piano and soloist, two pianos
For a concert balance in a studio with suitable acoustics it is possible
to get good separation by turning the soloist's microphone, which
may be bidirectional or hypercardioid, dead-side-on to the piano and
mixing in the output of a second and possibly a third microphone,
arranged to pick up a little of the piano. This time start by getting a
good sound from the soloist, then gradually fade in the piano, just
enough to bring it into the same perspective. It is of course also
possible to employ a backing track, as in popular music. The soloist
listens to pre-recorded piano on headphones, and the levels are
adjusted at the mixer.
A pianist who also speaks or sings may wish to use less voice than
would be possible without an additional vocal microphone: this
achieves a more intimate effect at the expense of internal balance.
132
Music balance
For the voice, a hypercardioid can be placed well forward, above or
slightly to the side, so that its dead angle (beginning 135 back from
the forward axis) is centred on the piano. However, any residual
piano pick-up may have an erratic frequency response, so try comparing this with an older style of balance that uses the smooth
bidirectional ribbon response. The ribbon is suspended about
50 cm (20 in) from the singer (as it would be for speech balance),
again with its dead side to the soundboard of the piano. It may be
directly in front of the singer and tilted down, or to the singer's right.
The ribbon is complemented by a wider (stereo) pair, which is faded
up just enough to establish piano stereo and natural reverberation.
Alternatively or additionally, add differential AR. Note that once the
faders are set, if one is moved separately from the other this can
change the balance radically.
[8.13] Balance for two pianos
Above: Mono, with two pianos on
a single microphone. Below: Two
pianos on individual mono
microphones (1 and 3) or on a
stereo pair (2).
The best way of balancing two pianos on a single microphone
depends on studio layout, on which the wishes of the pianists should
be respected if possible. With the pianos side by side, it is possible to
use a single ribbon or stereo pair at the tail. A two-microphone
balance might use two ribbons, each in the curve of one of the
pianos: these can be a shade closer than the single microphone, to
give a marginally greater degree of control over the relative volumes
and frequency pick-up.
Upright piano
[8.14] Upright piano
Position 1: Microphone above and
behind pianist's right shoulder.
A well-spaced pair might use
a microphone over each shoulder,
closer in if the front panel is removed.
Position 2: To right of pianist below
keyboard, or with a well-spaced pair,
one on each side. Position 3: At rear
of soundboard ± avoids action noise.
To avoid resonances at the back of an upright piano, move or angle it
away from any wall, then lift the lid and try the microphone somewhere on a line diagonally up from the pianist's right shoulder. Or
try two microphones, one on each side. If separation allows, try
taking off the front panel, then put microphones about 20 cm (8 in)
from the bass and top strings. Or take the lower panel off and again
use a pair, in this case spaced well apart to avoid foot-pedal noise.
Note also that because it is the soundboard that radiates, it is also
possible to stand a microphone at the back (for a close balance), or
diagonally up from the soundboard (for a more distant one). Behind
the soundboard there is less pedal action noise, but also, of course,
less brilliance and clarity.
One of these balances should also work for a janglebox, a type of
piano that is specially treated, with leaves of metal between hammer
and strings to give a tinny strike action, and with the two or three
strings for each note of the middle and top slightly out of tune with
each other. Listen and balance for the amount of action noise
required.
An upright piano with `electric' style loudspeakers may have them
fed from two very close microphones, that for bass clamped to the
frame near the top of Gii; the other somewhere between one or two
octaves above middle C (experiment for the best position). Where
separation is a problem use a contact microphone on the soundboard
Music balance
133
for the bass. The output may be taken to the mixer (with a feed back
to the two loudspeakers), or sent by DI box to the loudspeakers,
which are then balanced as for the guitar. Equalization (EQ) may be
used, either to emphasize high-frequency percussive transients or to
add weight to the bass.
Harp, harpsichord, celesta
[8.15] Harp
1, Frontal coverage, looking down to
soundbox. 2, Rear position for
microphone favours strings and
discriminates against pedal noise, but
also loses reinforcement from the
soundbox. 3, Close on soundbox,
can be used for light or pop music.
The harp has a soundbox below the strings: in performance this is
cradled by the player who reaches round on both sides to pluck (and
otherwise excite) the strings. Owing to the relative inefficiency of
the soundbox, the vibration of the strings contributes a higher proportion of the sound than in a piano, and ideally any balance should
cover both. Foot-pedals control the key of a harp, and if they are
operated during quiet passages, their mechanical sounds may be
audible. In an orchestra, harps generally benefit by spotting for
presence and by locating them clearly, usually towards the side of
the stereo picture.
A microphone about 1.5 m (4±5 ft) diagonally upward from the
soundbox may be directed to cover both that and the strings.
A ribbon gives tight coverage of one or both harps on its relatively
narrow forward lobe, while the rear lobe points up and away into the
hall. A condenser cardioid also covers the harps well, while a
hypercardioid can be placed closer in to the strings, favouring the
middle and upper parts of the soundbox, with some discrimination
against pedals.
A harpsichord has many of the qualities of the piano, including
mechanical action noise, plus the characteristic attack on each note
as the string is plucked by its quill. When the instrument is featured,
it is useful to have control over the relative values of transients and
resonance. For this, set out two microphones, one above and one below
the level of the soundboard. The lower microphone takes the body of
the sound, perhaps by reflection from the floor; the other, angled to
`see' down into the action, picks up the full sound but with transients
and upper harmonics somewhat over-represented. Mix to taste.
[8.16] Celesta
A microphone near but behind the
soundboard avoids mechanical noise,
which can be loud compared with the
music from this relatively quiet
instrument. Another possible position
is in front of the instrument, below
the keyboard, on the treble side of
the player's feet at the pedals, but
keeping well away from them.
In a baroque orchestra (covered by a coincident pair) harpsichord
continuo needs to be positioned well back near the basses, with
which it co-operates, and also to avoid its penetrating sound appearing too dominant. Use a spotting microphone to fine-tune its
contribution.
In other instruments of the same general class, such as virginals,
clavichord or spinet, apply the same principles: first identify the
soundbox or board, sources of action noise, attack transients and
higher frequency components, then place the microphone or microphones accordingly.
The celesta is a rather quiet keyboard instrument. In an orchestra it
benefits by spotting. Ideally, a close balance is needed for good
134
Music balance
separation, but this has the disadvantage of accentuating the already
potentially intrusive action noise. A balance at about 2 m (6±8 ft)
would avoid the worst of this, but if a closer balance is unavoidable
try a microphone either in the middle at the back or below the
keyboard on the treble side of the player's feet ± and keeping well
away from them.
Piano accordion
In a close balance the piano accordion may require two microphones, one directed towards the treble (keyboard) end, the other
favouring the bass. A closer balance on the air-pumping bass end can
be achieved by clipping a goose-neck electret to the instrument
itself, so that the microphone hangs over some of the sound holes.
For greater mobility, attach two, one at each end.
Harmonica
[8.17] Piano accordion
A popular music balance can be
achieved using close microphones at
the bass and treble sides. These may
both be attached to the body of the
instrument or just one attached over
the bass, with a stand microphone for
the treble. In this example, a special
plate and clip are shown, which can
also be used for guitar `amps'
(loudspeakers).
This can be something of an oddity: some players supply their own
specialized small microphone, which fits inside the resonant cavity
formed by their hands. This avoids the conventional alternative: a
simple stand microphone in front of the player.
Woodwind
Each of the wind instruments has a distinctive directional pattern of
radiation. In contrast to brass, on woodwind with fingered or keyed
holes the bell may make little difference to the sound: for all but the
highest frequencies, the main radiation is through the first few open
holes. In general, a microphone in front of and above the player and
directed towards the middle of the instrument (but not the player's
mouth) will prove satisfactory.
For wind, as with other sections of the orchestra, transients (the way
notes start) are of great importance in defining the instrument and
giving it character. These sounds must be clear but not exaggerated.
[8.18] Flute
A microphone behind the player's
head discriminates against the windy
edge tone, but can `see' the pipe and
finger holes.
On the flute, there is a breathy edge tone. The closer the balance, the
stronger are the transients in comparison with the main body of the
sound. Listening at an ordinary concert distance, the harsh edges of
the notes are rounded off; listening to a flute at 2 feet, it can sound
like a wind instrument with a vengeance. For classical music, `close'
can mean 1±2.5 m (3±8 ft); in a multimicrophone balance it may be
15±60 cm (6±24 in). In close balances it is occasionally advisable to
resort to devices such as placing the microphone behind a flautist's
Music balance
135
head. This reduces the high-frequency mouth noise and edge tones
without affecting the tonal sound radiating from the finger holes.
[8.19] `Electric' flute
Pick-up within the head of the flute
discriminates strongly against breath
noise, and offers a distinctively pure,
mellow tone.
At the other acoustic extreme is a headset carrying a miniature
microphone with its axis directed past the embouchure towards the
upper holes. Another possibility is an electret on a pad, taped in place
near the mouthpiece, but directed away from it. Taking this a stage
further, some flutes have been adapted so that the microphone is
actually inside the head joint, from where it picks up the sound wave
within the length of the flute, but hears no mouth noise or sound
radiated from the holes. This gives it the unnaturally pure tonal quality
used for the `Pink Panther' theme. Radiated sound can be added to
this as desired, from a stand microphone in front of the player.
An arrangement like that would not work for an `electric' recorder or
penny whistle, as the mouth covers the top end, but a microphone
can be fixed in to the side of the instrument above the finger holes.
When used for popular music the clarinet has been balanced using
two close microphones. One is on a stand at 15±30 cm (6±12 in),
pointing down towards the keys. The other is a lightweight electret
on a spring clip attached to the bell, angled by means of a short gooseneck to point back into it: this is mainly for frequencies over 3 kHz.
[8.20] `Electric' penny whistle
Pick-up can only be from the side of
the body of the instrument.
In a close balance of a full orchestra the woodwind may sound too far
back. To correct for this, sling an additional coincident pair or set in
three separate microphones on stands at a height of about 2 m (6 ft or
so) in front of the section. The extra microphones are for presence,
perhaps emphasized by presence in the 2±4 kHz range. They are not
intended to provide the main balance, and their signal must be added to
it with discretion. But they do also allow precise control of the stereo
spread in the central region, perhaps broadening it a little. The width
required from a stereo pair on woodwind would still, of course, be
much narrower than that of the main pair covering a whole orchestra.
Woodwind orchestral spotting is one of the cases for which a delay
may be calculated and added at the console. (Remember: delay in
seconds equals number of metres divided by 344.)
Saxophones
[8.21] Clarinet
Popular music balance using two
microphones, one of them on a
miniature goose-neck for the bell.
For a saxophone try a microphone near the middle keys, angled to
include some pick-up from the bell (for higher frequencies), but
listen also to check that the breath effects and keypad noises that
are also part of its character are present at acceptable levels. For a
breathy subtone quality, another pick-up angle might be along the
central axis of the bell. As a second microphone, an electret can be
attached to the bell, taping it on a soft pad at the rim, or by clipping
a goose-neck to the rim with the electret angled into the bell.
Saxophones may benefit from presence at about 2.5±3 kHz.
In big-band music a saxophone section has perhaps five musicians.
A balance with a bidirectional ribbon can have three players on one
136
Music balance
side and two on the other, with the microphone in line with the bells
and the lowest saxophone (usually a baritone) at one end of the
three. Players in this section may double on flute or clarinet, also
played to the same microphone. Another arrangement has the same
grouping, but with two hypercardioids angled down at the centre of
each side.
[8.22] Woodwind orchestral layout
F, Flute, doubling P, Piccolo. O, Oboe.
C, Clarinet. B, Bassoon. If the main
orchestral microphone is set for
a close balance, it may tend to favour
the strings at the expense of
woodwind, so that a spotting
microphone (or coincident or spaced
pair) may be needed.
An alternative that some balancers prefer is to place this section in a
line or arc, with one microphone to each pair of musicians. This
helps when saxophone players doubling woodwind fail to move in
for a second, quieter instrument, and also allows the height of each
microphone to be adjusted more precisely to the needs of the other
instrument. A close balance on clarinet, for example, would be lower
than for a flute.
Brass
In brass, the axis of the bell of each instrument carries the main stream
of high harmonics. Even when the music requires that this be directed
at the audience, only a relatively small group receives the full blast at
any one moment, the rest hearing the sound off-axis. In consequence,
for all but the most exaggeratedly brilliant effects, the main microphone for an orchestral balance is usually quite satisfactory for the
brass. It is no disadvantage that its distance from the microphone is
greater than that of some other sections, because the added reverberation conventionally suggests the sheer volume produced. Orchestral
brass that is balanced well back in a spacious hall is usually easier to
control (requiring relatively little manipulation of level) than when, by
choice or necessity, closer microphones are used.
Spotting microphones may have to be set for an unusual reason: in
this case because the volume of sound from brass can cause the
whole balance to be taken back, making this section seem even more
distant unless there is separate control of presence ± which will
require additional reverberation to match its perspective.
[8.23] Saxophones in dance band
In a studio layout, up to five players
may be grouped around a central
microphone. Above: A bidirectional
microphone set at the height of the
instrument bells. Below: C, Two
hypercardioid microphones set above
bell level and angled downwards to
reduce pick-up from outside this area.
S, Solo microphone for player
doubling on a second instrument.
In a close balance ± with a trumpet at 30 cm (1 ft) or so, or trombone
closer than that ± the volume from brass can reach 130 dB, higher
than the maximum recommended for the condenser microphones
needed to get the full frequency response, unless the signal can be
reduced by switching at the microphone itself, for which a pad of up
to 20 dB is required. Do not rely on rehearsal levels of unknown
brass players: instead, allow for as much as a 10 dB rise in level
between rehearsal and recording, especially for a live broadcast,
otherwise there is a risk of distortion, reduction of the working range
on the faders and loss of separation. It also helps if the players are
prepared to co-operate by leaning in to a microphone on quiet
passages with mutes, then sit back for louder passages, and also
play far enough to the side of it to reduce overload and blasting. The
microphone must be near enough for them to reach for close work
but far enough back to pick up several players equally. Brass players
Music balance
137
are well aware of the directional properties of their instruments and
will play directly towards the microphone or not as the music
requires, though some bands that have done little work in studios
and more in dance halls may take this to extremes, perhaps by
playing soft passages in to a music stand. They may need to be
reminded that a cleaner sound can be achieved if some of the control
is left to the balancer.
The microphone should be at a height that is close to the axis of the
instruments as they can be comfortably played ± although with
trombones this may be awkward, as music stands can get in the
way of the ideal position. However, the leader of the section will
usually state a preference for whether microphones should be above
or below the stand: this probably depends on how well the players
know the music. In a band where the music played is part of a limited
and well-rehearsed repertoire, trombones are often held up, but
musicians faced with unfamiliar music need to look down to see it.
Miniature goose-neck electrets can also be clipped to the rim of the
bell of a brass instrument, a technique that gives the player freedom
of movement, but loses the effect of movement on performance.
Again, try different microphone positions by bending the goose-neck.
Brass may benefit from the presence of 5 dB or even more at about
6±9 kHz. But note that some microphones have peaks in their
response in this region and produce the effect without much need
for equalization.
In the orchestra, horns can often be placed where their sound reflects
from a wall or other surfaces behind them. Such an arrangement ±
with a microphone looking down from above at the front ± is suitable
in most circumstances; though exceptionally, for greater separation
and a strong, rather uncharacteristic quality, a microphone can be
placed behind the player in line with the bell.
To separate a tuba, try a microphone at 60 cm (2 ft), slightly off-axis.
Several layouts are used for brass or military bands. In one, sections
form an arc round the conductor much as in a string orchestra; in
another, the performers make three sides of a hollow square, with the
cornets and trombones facing each other on opposite sides. In this way
the tone of the horns and euphoniums at the back is not blotted out by
the full blast of more penetrating instruments. A microphone pair may
be placed a little way back from the square, as for an orchestra.
[8.24] Miniature goose-neck
Electret for brass (trumpet,
trombone, tuba and also saxophone)
in multimicrophone balances,
allowing player freedom of
movement.
The main body of euphoniums and bass tubas radiates upward and can
sound woolly on the main microphones unless there is a reflecting
surface above them to carry their sound forward. Failing this, an
additional microphone above them will add presence. A solo euphonium player should come forward to a microphone set high up and
angled to separate it from other instruments of similar sound quality.
A solo cornet may also have its own microphone, a condenser cardioid.
A military band behaves like a small orchestra from which the strings
have been omitted. There is usually plenty of percussion, including
such melodic instruments as the xylophone and glockenspiel, and
138
Music balance
care must be taken to ensure that these are given full value in the
balance.
[8.25] Drum microphone
A microphone a few centimetres
inside the rim is suitable for most
of the smaller drums in a kit.
A military band in marching order is balanced for the benefit of the
troops behind them; in the open air, woodwind, at the back, will not be
reinforced as it would indoors. A microphone balance from the rear
helps to correct this, even though for television or film, pictures are
more likely to be from the front. A band on the move is even trickier,
as the best balance will rarely match preferred wide-angle camera
positions. A marching band in an arena can be followed well with a
tetrahedral (Soundfield) array: suspended over the action, it can be
steered to give an angle and polar response that covers the band, while
discriminating against excessive reverberation or crowd noise.
Percussion, drums
Most percussion instruments can be recorded with ease. However,
some percussion sounds register more easily than others, and on all
but the most distant microphone set-up it is as well to check the
balance of each separate item with care. Any rostra (risers) need to
be reinforced to prevent coloration. In orchestral percussion a microphone with an extended, smooth high-frequency response is required
mainly for triangle, cymbals and gong. Spotting microphones may
be set for many of the individual components: for the quieter ones,
so that detail is not lost; and for the louder (as for brass), so that
presence can be retained when overall volume is reduced. In an
orchestral balance delay is often used with spotting microphones.
For instruments such as the glockenspiel and xylophone use a stereo
coincident pair above them (though some balancers are happy with
a well-separated spaced pair). The characteristic sound of the
vibraphone requires a close microphone or microphones beneath
the bars, angled upward to the tops of the tubular resonators (the
bottoms are closed).
In popular music, percussion shows itself mainly in the form of the
drum-kit, including a snare drum, bass drum (foot-pedal operated, as
a kick drum), a cymbal (or several), a hi-hat (a foot-operated double
cymbal), and toms in different sizes.
[8.26] Drum-kit
1, Kick drum. 2, Cymbals. 3, Toms
(and other drums). 4, Snare drum.
5, Hi-hat. 6, Coincident pair with good
high-frequency response (a spaced
pair may be used instead).
7, Additional microphone (moving coil)
for percussive kick-drum effect.
Further microphones are usually
added, to give each drum and the
hi-hat its own close pick-up.
Music balance
139
In the simplest balance, a microphone pair (with an extended highfrequency response) points in from above. This will certainly cover
the snare drum and cymbals, but any attempt to move down for
a closer balance increases their dominance over the hi-hat, which
radiates most strongly in a horizontal plane (in contrast to the
up-and-down figure-of-eight pattern of the top cymbals), and the
kick drum. Avoid a position alongside the hi-hat, as it puffs out
sharp gusts of air.
For a tighter sound move one microphone closer to the hi-hat, also
favouring the left-hand top cymbal or cymbals and some of the snare
and small tom-tom, together inevitably with some bass drum; a
second cardioid microphone is then moved in over the other side
of the bass drum to find the other cymbal or cymbals, hi-hat, big
tom-tom, and again, perhaps, the snare drum. Changes of position,
angle, polar diagram or relative level can give reasonable cover of
nearly every element except the kick drum.
The distinctive kick-drum sound that is required to fill in the texture of
much popular music can be obtained by facing the diaphragm of a
moving-coil microphone toward the front skin and close to its edge,
where a wider and more balanced range of overtones is present than at
the centre. This is one place where one of the older, more solid-looking
microphones with a limited top response is as good as a modern highquality microphone (which might in any case be overloaded). For
the dry thud required in pop music put a blanket inside against the
beater head to shorten its vibration. Try using EQ to reduce the lower
300±600 Hz range and lift at 2.5±5 kHz to make the sound crisper.
For greater control still (and to increase separation from cymbals) use
separate microphones for each part of the kit. For some of the drums
(snare, toms, congas, bongos, etc.) they may be clipped to the top edge
(tensioning hoop) or pointed down from stands, 3±8 cm (1±3 in) over
the head and 5 cm (2 in) in from the rim. A relatively insensitive electret
can also go inside a kick drum, but this needs a solid mounting. With a
low stand microphone for the hi-hat, an overhead pair for the cymbals
and basic stereo coverage, the drum-kit alone can use up far more
tracks in a multitrack recording than any other part of a group. Many
pop balancers work with the drummer to tailor the sound, and in
particular modify the snare drum by tightening snare tension (perhaps
also putting some masking tape on it), and may tape a light duster
over part of the skin (away from where it is struck). Presence
(at 2.8 kHz or more) plus additional bass and extreme top may be added.
[8.27] Miniature microphone for
drums
1, Condenser hypercardioid, clamped
on top hoop or tensioner: the tone
varies with drum balance. 2, The
microphone on a small stand inside
the kick-drum shell, fairly close to the
skin. If a blanket is also put inside to
press against the skin, it shortens
the vibration to give the drumbeat
more attack.
The toms are less likely to require attention, but the quality of
cymbals varies and the application of masking tape may again help
to reduce the blemishes of poorer instruments. With such extreme
close balancing it is wise to check the maximum playing volume in
advance, and also to recheck the quality of the sound from each
drum from time to time: it may change as a session progresses.
Stereo width is introduced by steering the various microphones to
slightly separate positions. If a contribution from the studio acoustics
is required, angle in two ribbons from perhaps 1.5 m (5 ft) above and
away from the kit.
140
Music balance
Singers: solo and chorus
What counts as a close balance for a singer depends on the type of
music. For a pop singer it may be close to the lips, for an operatic
aria at arm's length. For classical music a clean, flat response is all
that is necessary: try a hypercardioid condenser microphone at about
1 m (3±4 ft), angled up enough to point over any orchestral instruments
behind the singer. High-quality ribbons at the same distance can also
work, though with care for what might lie in the opposite lobe.
[8.28] Headset (without
headphones)
For singers, designed for light weight
and minimal disturbance of hairstyle.
The electret microphone is in a fixed
position, and at this distance must be
beyond the corner of the mouth in
order to avoid popping.
For lighter styles of music using a close balance, compensation for
bass tip-up (proximity effect) is, as ever, needed when using a fully
directional microphone, and if the singer moves back, has to be
restored. More useful are directional microphones on which the
low-frequency response degenerates towards omni. But while a
microphone that is hypercardioid only at high frequencies discriminates well against the kind of high-frequency background that might
interfere with melodic line and intelligibility, it still picks up a
boomy rumble, so again the bass must be reduced ± though a singer
working very close does produce less of the chest tone that in opera
is used to deepen and strengthen the sound.
A skilled performer may use the angle and distance of a directional
microphone to precise effect; less reliable singers are certainly safer
with an omni at lower frequencies, for which only the volume has to
be adjusted (though even this may have some variation of frequency
response with angle). For total freedom of movement (and free
hands) but constant distance and angle, try a headset (with headband
below the hair at the back, for those whose hairstyle is part of the
act). The tiny electret must be a little away from the corner of the
mouth in order to escape `popping'. For other microphones, popping
is minimized by directing the voice (as for speech) across the
microphone ± that is, by missing it with the air blast. Windshields,
built-in or added, also help.
To improve separation in a studio, a singer can be given a tent of
screens, open (for visual contact with any musical director) only on
the dead side of the microphone, or with a transparent panel on that
line. In some balances a vocal guide track is recorded (on a spare
track) at the same time as the main instrumental sound, and the lead
and backing vocals overdubbed to replay.
[8.29] Vocalist
When the volume of surrounding
instruments is high, the singer is
screened. The microphone's live side
is towards absorbers and its dead
side towards the open studio.
Some singers seem subject to a superstitious belief that close balance
gives them presence. In fact, if separation (the real reason for close
balance) can be achieved in other ways, try setting in a second
microphone 30±60 cm (1±2 ft) from the singer's mouth. Not only will
popping no longer be a problem, it may also be found that the more
distant balance is better, as well as easier to control. Presence can be
added electronically, at about 1.5±3 kHz or higher, depending on the
voice. Sibilance can be cured by some combination of singing across
the diaphragm, careful enunciation, high-frequency roll-off, and
choice of microphone with a smooth 3±12 kHz response. Condenser
cardioids are preferred for this quality rather than for their extended
Music balance
141
high-frequency response, which is not needed. Some balancers are
happy to use microphones with multiple high-frequency peaks.
A folk-singer with guitar can in principle be balanced on a single
microphone, perhaps favouring the voice, but two microphones
mounted on the same stand give greater control of relative level,
together with the capacity for separate equalization and artificial
reverberation. According to the 3:1 rule, the distance between each
microphone and its sound source must be less than a third of the
distance that separates them. Another way to avoid phase interference is to delay the vocal signal by about 1 ms, to match its time of
arrival at the guitar microphone.
[8.30] Singer with pop-shield
A gauze on a 10 cm (4 in) hoop
protects the microphone from the
vocal gusts that accompany a
plosive `p' sound.
Individual singers are rarely balanced in stereo (except where action
is taken from the stage). Mono microphones are used and steered to a
suitable position in an overall balance. But a coincident pair is well
suited to a backing group or chorus, which can then be given any
desired width and position in the final mix.
With choral groups, strive (generally) for clarity of diction. Set the
microphone back far enough to get a blended sound, but without
losing intelligibility. Spaced bidirectional or cardioid microphones
may cover a large chorus. Where a small group can conveniently
work to a single microphone position, use a cardioid or, for stereo
width, a coincident pair, curving the front line of singers to match
the microphone's field pattern.
There is no special virtue in clear separation of sopranos, altos,
tenors and basses into separate blocks: the stereo effect may be
good, but the musical quality of a layout that blends all sections
may be better. As an alternative to the block layout, a chorus may be
arranged in rows (and here even more than in the other cases the
microphones need to be high up in order to balance all the voices).
Occasionally, the singers may all be mixed together. Opera singers
must do this, and so can some choirs. Skilled musicianship is
required: when a group can sing like this and wishes to do so, the
balancer should not discourage them.
[8.31] Orchestral balance
1 and 2, Two stereo pairs with
different polar diagrams but giving the
same stage width to the orchestra
(3). Microphone 1 is a double
figure-of-eight, microphone 2 is a
double cardioid: a mixture of the two
can be used to control and change
reverberation where there is such
a wide dynamic range that much
volume compression is necessary.
4, For a soloist a spotting monophonic
microphone is used, steering this to
the same position where it appears
on the audio stage, as established
by the stereo pair.
Church music, and particularly antiphonal music with spaced choirs
of voices, benefits enormously by being brought round to the full
spread of stereo: it was here that surround sound and its predecessor,
quad sound, came into their own without the need for special
composition.
The orchestra
Our next individual `instrument' is the full orchestra. Indeed, it can
really be considered to be one instrument if it is well balanced
internally.
The detail of orchestral layout is not absolutely standardized, but the
strings are always spread across the full width of the stage, grouped
in sections around the conductor, with the woodwind behind them
142
Music balance
near the centre. In an older arrangement the violins were divided,
with first violins on the left and second on the right, then violas,
cellos and double basses in the arc behind them. For stereo, an
advantage of this is that the violin sections are clearly separated,
and the weight is in the middle. But having the upright instruments
(cellos and basses) face forward is not necessary, for either the
microphone or the audience, as they are virtually omnidirectional
radiators; and there is also some disadvantage in having
high-frequency radiation from second violins directed to the back,
away from both audience and microphone. The modern layout has
the sections placed, from the left: first violins, second violins,
violas, cellos; and raised behind the cellos on the right, the basses.
Woodwind is always across the centre, and percussion and timpani
at the back on one side, with brass around much of the back to the
other. That leaves horns, harps and possibly celesta, harpsichord,
orchestral piano, etc., all of which may appear in different places.
[8.32] Stereo orchestral layouts
1, With a spaced pair to `pin down'
the corners of the orchestra (e.g. harp
on left and basses on right). 2, With
a spaced pair used to increase
reverberation.
While the balancer's job is generally to reflect what is offered, in the
studio a well-established and experienced balancer may also offer
suggestions for improvement, often through a producer. A conductor
who is thorough may consult the balancer directly, not only on
matters of detail in placing, but might also ask how the internal
balance sounds. Many conductors, however, concentrate solely on
musical qualities, so that if, for example, there is too much brass it
will be left to the balancer to sort it out, if that is possible.
In an orchestral balance that is based on coincident cardioids, theory
suggests that the angle between them should be 131 . Treat that as a
starting point. Try placing them first to control width, rather than
reverberation. This means that for a subject that is to occupy the
major part of the audio stage the microphones will look down from
the edge of the pick-up area; for an orchestra that is to occupy about
two-thirds to three-quarters of the sound stage, they are as high but
set back a little from the conductor. Then listen to compare direct
sound (enriched by early reflections) with reverberation: this offers a
subjective approach to microphone placement relative to critical
distance. You can always add reverberation but not take it away.
Reverberation may be added by a variety of means. One is to have a
second coincident pair set back from the first and facing toward the
empty part of the hall or studio. Another has a spaced pair of
cardioids at the sides of the hall, well forward again, but looking
back along it. In either of these cases the two microphones must be
steered to their correct sides of the studio, as some direct sound will
reach them, and even more by short-path first reflections. Reverberation microphones should not be more than 12 m (40 ft) back, as this
introduces echoes, most noticeable on percussive sounds such as
drums. Alternatively, resynchronize by using delay on the closer,
main microphones.
[8.33] Balance for 5.1 surround
sound
A stereo pair with omnidirectional
microphones 1.2 m (4 ft) on either
flank and others set back 9±12 m
(30±40 ft).
Balances using these combinations of forward and reverberation
microphones have been used for 5.1 surround sound. A wide range
of other successful 5.1 balances have been reported, but some may
work best in their original locations or when operated by their
originators. The hall's acoustics can easily make more difference
Music balance
143
to the result. In practice, any balance that offers good stereo (clean
positioning and a well-judged compromise between richness and
clarity), together with separate reverberation coverage, is likely to
work well in 5.1.
Another stereo layout has a second coincident pair of microphones
for reverberation, but this time figures-of-eight looking forward and
set back to such a distance that the positions of the various parts of
the image roughly coincide. For an orchestra, the second pair may be
about 3 m (10 ft) further back. This arrangement has the additional
advantages that the balancer can work between the two pairs to
handle extreme changes of volume, and that the second pair gives
an automatic standby facility in the event of failure or loss of quality
of the first pair on a live broadcast. Such failure is rare, but the
facility is certainly desirable, and in any case the two pairs of
microphones can be used for balance comparison tests. Phase
cancellation on the back of the second pair is still a potential problem.
[8.34] Reverberation in stereo
orchestral layouts
1, Good: behind the main pair is a
second coincident pair with reversed
output, to control reverberation
separately ± this can also be used for
surround sound. 2, Bad: a forwardfacing spaced pair set back into the
hall may give false information from
the brass. Directed towards the
microphone on the opposite side, it
may sound as though the brass is in
the wrong place.
[8.35] Microphone `tree'
1±3, A narrowly spaced pair of
microphones, with a third forward
from the middle, in a balance
originally developed by a recording
company. A surround version of this
has additional microphones at the
sides: these are mainly for
reverberation from the hall, but also
broaden the orchestral image a little.
4, In yet another balance, a baffled
omni pair (similar to the `dummy
head' layout) has been favoured for
its spaced effect.
Any microphone that is set up in addition to the main pair and facing
toward the orchestra (or a `bright' reflecting surface) may give false
information about the position of the brass, if significant direct or
strong first-reflection sound can reach it. The trumpets are likely to
be directed over the heads of the strings and may be picked up on
any microphone in their line of fire, even on the opposite side of the
studio. There is then an extra trumpet signal, pulling the section
across the sound stage.
Further mono microphones are often added (for 5.1 too). For stereo,
one on each side of the orchestra may be placed to open out the
strings. These are faded up just sufficiently to lend presence to string
tone, and panned to their proper places in the stereo balance, which
is thereby more clearly defined. Woodwind is also often covered:
where the main pair give good basic stereo coverage a mono microphone may be sufficient, though a stereo pair is often used here too.
It lends presence to woodwind solo passages. Harps and some other
instruments may also be given spotting microphones, which again
must be used with subtlety to add clarity and positional information.
When close spotting microphones are mixed into a relatively distant
overall balance like this, check whether they seem to bring their
instrument or section too close. If the added volume is low and the
difference in distance not too great, this effect may not be noticeable. If in doubt, it may be safer to insert delay in the spotting
channels (the number of seconds delay is distance in metres divided
by 344).
Spotting microphones and delay can also be used to restore the
balance of sections that are very weak in numbers. Automatic double
tracking (ADT; also available on many consoles) adds a copy of the
original signal with a delay that can be set somewhere between 4 and
20 ms: this doubles the apparent number of musicians.
With spaced microphones, observe the `3:1' rule for separation.
When using pairs of omnidirectional microphones, for each unit of
distance between the two microphones, allow no more than a third of
that distance between each microphone and its subject. This ensures
144
Music balance
[8.36] The 3:1 rule
In a multimicrophone balance the
distance between two
omnidirectional microphones should
be at least three times the distances
from each microphone to its sound
source. This ensures that separation
is greater than 10 dB.
a separation of at least 10 dB, to keep phase-cancellation problems at
bay. With directional microphones, this rule can be relaxed a little,
but with caution.
Orchestra: spaced microphones
This book recommends coincident pairs for stereo, but accepts that
spaced microphones have always had strong advocates, and that the
availability of omnidirectional microphones with an exceptionally
smooth response reinforces their case. One such arrangement is
outlined below.
[8.37] Orchestral multimicrophone
balance
1, Main stereo pair. 2, Woodwind pair
2±3 m (7±10 ft) forward of this
section. 3, 4, Spotting microphones
for string sections, about 2 m (6 ft)
from them, add presence and
strengthen positional information.
5±9, A curtain of omnidirectional
microphones. Phase and control
problems are reduced by a
sufficiently close balance. Once
faders are set, these are controlled
as a group, because changes in the
relative level of nearby microphones
can affect the apparent position of
instruments equidistant from them,
including loud, distant sources such
as brass. 10±12, Omnidirectional
`drop' microphones, mainly for
woodwind. 13, 14, Spaced bar of
omnidirectional microphones used
as an alternative to the curtain.
15, 16, Spaced reverberation
microphones.
A `bar' with two omnidirectional microphones separated by about
40 cm (16 in) gives a rich, spacious sound. Their faders must be
locked rigidly together, as the slightest difference swings the orchestra sharply to one side. The axes are angled slightly outwards to
allow for high-frequency reinforcement normal to the diaphragm. In
a concert hall, their distance may not be easy to change, so a curtain
of some five extra microphones may be added, close to the front line
of the orchestra, with three more set deeper to favour the woodwind.
Their outputs are all panned to their appropriate positions within the
stereo picture, and after their relative levels have been set by listening to the effect of each in turn, they are controlled as a group,
adding just enough from them for presence and positional information (rather weak from the space-bar) to be strengthened. As usual,
fade up until `too much' is heard, then set them down a little.
Again, spotting microphones (as well as those for woodwind) are
added, and considered as candidates for EQ and artificial reverberation. Digital delay of 10±16 ms can be used to avoid time
discrepancies.
The virtues of this mass of microphones are still debated. Where a
series of concerts is to be covered by several balancers it is best if the
hall is rigged in advance with all the main systems currently in
contention, so that direct comparison tests can be made. Critics of
space-bar and curtain techniques describe them as `muddy'; critics
of coincident pairs call them `clinical'.
Pipe organ with orchestra
If a pipe organ is used with orchestra there is a possibility (in stereo)
that its distance will cause its image to be too narrow.
Music balance
145
In fact, as a good organ is laid out in a hall, it should have some
width (indeed, it is bound to), but not so much that the spatial
separation of the various voices is marked: like a choir, organ tone
is better blended than broken up into obvious sections. However, in
stereo, extreme width (such as that of the organ at the Royal Festival
Hall in London) is narrowed by a distant coincident pair, and an
average width may even seem narrow. A second pair at a distance
that `sees' the organ as a whole can be used to broaden it on the
sound stage. They should be set high and angled to avoid the
orchestra, and the main pair set to discriminate against direct sound
from the organ.
[8.38] Organist on headphones
Where a keyboard is badly placed for
the organist to hear a good balance,
mixer output may be fed back on
headphones.
Orchestra with soloists or
chorus
Concert works with solo instruments near the centre can in principle
simply use a standard orchestral balance, as the main microphones
already favour the soloist. However, in most cases it is usual to set in
spotting microphones for additional presence. In contrast, for a piano
concerto it may be necessary to angle or place the main microphones
to discriminate a little against the volume of the solo instrument.
If there are several soloists, as in a double or triple concerto, or
several solo singers, they need to be placed with a thought for the
layout of the stereo sound stage.
In public performances a chorus is usually placed behind the orchestra. This creates an imbalance that is greater at the main microphone
position than for most of the audience, which is farther back. If the
chorus is already weaker than the orchestra, this makes it worse. But
in any case, spaced directional microphones are usually added to
boost presence and intelligibility, together with extra volume if
needed. These are set on very high stands or suspended at the back
of the orchestra, to look down at the choir with their dead side to the
orchestra. These closer microphones may require delay, to match
the choir's distance on the main. Also, the stereo spread can be
increased if on the main microphone they appear to be bunched too
close together. Weak forces can then be augmented by automatic
double tracking (ADT).
For television, to clear sight lines, slung omnidirectional microphones are sometimes used. If so, these may require equalization
by adding sufficient lift at 3 kHz to restore intelligibility.
[8.39] Orchestra with choir
1, Main pair. 2, Choir pair.
3±5, Directional spotting
microphones. 6±8, Alternative layout
for choir. Since any spill onto a central
coincident pair will be reversed,
6 should be steered to the right and
7 to the left.
Opera
Opera may be recorded or broadcast in concert form; or it may be
taken from a public stage performance. A range of methods has been
used for each. The simplest employs a standard layout for orchestra,
146
Music balance
with chorus behind, and soloists at the front: this works well for
oratorio and for concert performances of opera where there is no
pretence at dramatic presentation.
[8.40] Stage stereo pair, set in
orchestra pit
Capsules on goose-necks are angled
slightly down, almost touching the
stage. These act as boundary
microphones with a directional
response parallel to the stage
surface, which should be clear of
absorbers for at least 1 m (3±4 ft).
Additional spaced microphones will
be required to cover action out to the
sides, or which approaches too close
to the central pair.
Opera on records may have stereo movement, which can be introduced both laterally and in perspective. One layout used for this had
a line of five cardioid microphones beyond the orchestra: these
covered an acting area marked in a checkered pattern. The square
from which to perform was marked on each singer's score; moves
were similarly indicated. A disadvantage was that the conductor
could not hear the soloists as well as with an oratorio layout. An
alternative had singers on either side of the conductor, but facing the
orchestra and working to cardioids, which therefore had their dead
sides to the orchestra. The singers were recorded on separate tracks,
and their perspective and position (with movement provided by
panning) were arranged at the mix-down stage. For this, less preparation was required in advance.
In an unusual layout that makes good use of studio space, a standard
orchestral layout is combined with a laterally reversed row of singers. With the orchestra at one end of the studio, the chorus is placed
at an angle to it along one side of the studio, so that a corner of the
orchestra is next to the end of the chorus. The chorus, in rows on
rising rostra, works to a line of microphones, as do the soloists in
front of them. These microphones are then `folded' back across the
orchestral sound image.
Opera from the stage can be covered by a coincident pair for the
orchestra and a row of cardioids along the front of the stage for the
singers. If the main pair has to be raised to clear audience sight lines,
close microphones are required in the orchestra pit.
To cover the stage, there are further possibilities. One is a coincident
pair of cardioids (at 130 ), downstage centre, with a spaced pair at
the sides, to fill in the corners. If these are all are mounted on
brackets, with diaphragms low and close to the edge of the stage,
sound reflections from the stage are minimized. The main pair also
picks up some sound (though not a full balance) from the orchestra.
[8.41] Microphone positions in the
theatre (e.g. for opera)
1 and 2, A pair of (cardioid)
microphones in the footlights.
Though very close to the orchestra,
their response discriminates
sufficiently against this. 3, A single
microphone or coincident pair placed
high in the auditorium ± but well
forward. At this distance, a cardioid
response again favours singers.
A similar arrangement employs directional boundary microphones on
the stage. Again, there may be a coincident pair at the centre and
single microphones at the corners, but on a wide stage three separate
pairs can also work well: the width of each pair is reduced, with the
inner microphone of each outer pair steered to the same position as the
corresponding element of the central pair. This gives smooth coverage
of movement across the stage. If there is a central prompt box to get in
the way, two pairs might serve. Note, however, that if boundary
microphones are placed on a flexible rather than solid surface, their
directional response can change in ways not predicted by simple
theory, perhaps by picking up more orchestra than expected.
A deficiency of any footlight microphones is that they can all too
easily pick up footsteps on the stage. To counter this, if feet
approach any microphone too heavily, fade that one out and
Music balance
147
rebalance on another to one side. Check the effect on the stereo
picture: although displaced, this may be acceptable, or it may be
better to pan the remaining microphone to restore the image to its
former position.
Important action in an upstage area that is not covered adequately on
downstage microphones can perhaps be taken on a `drop' microphone
(i.e. suspended by its cable). Take care not to overdo its use, bringing
intentionally distant scenes too far forward. Balance with an ear to the
clarity of vocal line, even when television subtitles are added.
[8.42] Studio opera layout with
chorus and soloists reversed left to
right and superimposed on stage
A converted theatre is used in this
example. 1, Proscenium arch: the
stage is extended into the auditorium.
2, Soloists' microphones. 3, Chorus.
4, Conductor. 5, Announcer. This
microphone, dead side to the
orchestra, can also be used for
spotting effects (e.g. castanets,
steered to represent the movement of
a dance). 6, An offstage (i.e. distant)
chorus can be placed underneath the
theatre balcony without a microphone.
7, Audience above. Orchestral layout:
V1, 1st violins. V2, 2nd violins.
VA, Violas. C, Cellos. D, Basses.
H, Horns (with reflecting screens
behind).F,Flutes.O,Oboes.C,Clarinets.
B,Bassoons.T1,Trumpets.
T2,Trombones.P1,Timpani.P2,Other
percussion. Hp, Harp. Note that the
harp is `in' or near the right-hand
speaker, and that the left-hand side of
the orchestra and the side of the
chorus and soloists nearest to it will be
picked up on each other's
microphones, and must therefore be
arranged to coincide in the audio stage.
For the singers, examples of
positioning for particular scenes using
this layout are as follows: (i) Part of
chorus left, principals centre, off-stage
chorus right. (ii) Duet at centre solo
microphones, harp figures right.
(iii) Quartet at four centre solo
microphones.
Classical music in vision
For classical music, a television picture makes only a little difference to the type of balance that is sought. But getting a good balance
of any kind can be a little more complicated, because:
.
television studios are generally too dead ± the halls that are normally
used for concerts (or even music sound studios) are therefore to be
preferred in cases where they can conveniently be used;
148
Music balance
.
.
microphones should not be obtrusive;
the layout must be visually satisfying, although a good rule is
to aim for a layout as close as possible to that of a concert
performance.
What is not needed is any attempt to make sound follow picture: when
the camera zooms in to a close-up of the oboe, the sound should
remain at the concert perspective (note that this is the exact reverse of
what may be done with speech). The big close-up in sound would be
permitted only for a special effect; if, for instance, someone is talking
about the role of the oboe in the ensemble, or alternatively, if this is a
dramatic presentation and we must hear as we see the action, from a
position within the orchestra. Even for these, the overall orchestral
sound would be distorted as little as possible.
One slight difference in balance is desirable for television: it helps to
have marginally greater clarity at the expense of the tonal blending
that many people prefer on radio and records. The extra presence
permits individual instruments or groups to be heard just clearly
enough for close shots to be aurally justified. This does not necessarily
mean shorter reverberation, but simply that the attack on each note
must be heard clearly: this is usually enough to identify individual
instruments. A technique that can be used to create this effect is to
mix the outputs from two main microphone positions, one closer and
the other more distant than for the usual balance.
A balance with this extra clarity also takes into account the different
listening conditions that may exist in the home: the sound quality
of loudspeakers on many television receivers is lower than that used
for radio and records by many people who enjoy music; a more
transparent balance serves this better.
For ballet performed in the television studio it is often difficult to get
both the dance settings and the orchestra into the same studio: this
may be a blessing in disguise, however, as the solution is to take the
orchestra away to a music studio (with remote cameras, if required).
The music can then be replayed to the dancers over loudspeakers.
[8.43] Ballet: playback and effects
Dancers need the music to be fairly
close, so a loudspeaker may be
necessary. Here a series of
loudspeakers set in a vertical line
radiates high frequencies in a
horizontal field. A gun microphone
above them receives only low-pitched
sound from this array, and this can be
filtered out, as the main component
of the dance effects is at higher
frequencies.
Limitations (due to cost) on orchestral rehearsal time can mean that
orchestra and dancers must be rehearsed separately. If, in addition,
the orchestra is pre-recorded, the tempi are then set, and the dancers
know what to expect; ideally, principal dancers will have attended
the recording. If piano transcriptions are used for rehearsal, it may be
a responsibility of the sound department to make and replay a
recording of this. The dancers must have clear audio replay at all
places where the action takes them, so the action studio must be
liberally supplied with loudspeakers. In some cases in ballet the
dancers start before the music; where this happens the replay operator must be as precise and accurate as an orchestral conductor in
timing the start of music to match the action.
Effects microphones ± usually parabolic or gun microphones, or
hypercardioids angled to discriminate against loudspeakers ± are
needed in the studio to add sufficient noise of movement (the
dancers' steps, rustle of costume, etc.) to add conviction to the
Music balance
149
pictures of the dance. Consider also an adapted-array (digitally
processed) combination microphone if there is one available: its
8 30 pick-up lobe should cover the action.
For singers, the type of balance depends on whether a microphone
may appear in vision; if not, a boom might be used. With the more
distant balance, check for separation problems due to orchestral spill
on the singer's microphone. Condenser cardioid microphones are
preferred.
A marginal advantage of the relative deadness of television generalpurpose studios is that when AR is used it can be added differentially
± not just in different proportions, but also with different reverberation times for different parts of the sound. Singers usually need less
than that required for the music that accompanies them.
[8.44] Studio communications
Links between a television studio and
(in this case) orchestral sound. The
singers' voices are fed back to the
conductor.
Vocal dynamics are an important part of any singer's performance,
but when this is broadcast, some compression may be necessary, and
in television the problem is made more complex by movement of
both singer and microphone, and also by any variations in distance
between singer and microphone to accommodate the frame size.
Here, even more responsibility for interpretation of the artist's role
may fall to the sound balancer, who must therefore think not only
about problems of coverage but also ± along with the director and
singer ± about the overall artistic intention.
Where singers and accompaniment have to be physically separated,
time-lag problems may need to be rectified. Foldback loudspeakers
are needed to relay sound to the singer; the conductor, too, needs to
hear how the singer's contribution fits into the ensemble.
For some wide shots, pre-recordings may be used: timing these
to match live action requires planning and rehearsal. An additional recorded track can be used for cues that are fed only to the
performers. Remember, always, that foldback loudspeakers may
be safer when placed off the rear axis of a directional microphone.
Televised opera
There are several ways of presenting opera on television:
.
.
.
.
from the stage;
with actors (perhaps including some of the singers) miming to
pre-recorded music;
with singers in a television studio and orchestra in a sound studio;
with both singers and orchestra in some hall of suitable (large)
size and good musical acoustics.
For stage performances, use the balance already described. If possible, work from a separate sound control vehicle with good loudspeakers, a reasonable volume of space and suitable acoustic
treatment.
150
Music balance
[8.45] Opera or ballet in the
television studio
Layout for conductor with orchestra
in a separate sound studio. 1, Line
loudspeaker along top edge of music
stand. 2, (Alternative) hood
loudspeaker over conductor's head.
3, Monitor for conductor to see
action. 4, Camera to relay beat to
repetiteur.
Miming opera was once extensively used in continental Europe. In
Britain it has been used less, in the belief that the singer's full
performance has greater musical conviction. One arrangement
employs separate studios. That for the orchestra has good musical
acoustics; the action studio has all of the regular facilities to cover
sound and pictures ± for which the original grand, and deliberately
unrealistic, style of stage opera may be modified to that of music
drama. The singers can therefore be covered by booms (and, where
necessary, other microphones) in a way similar to that for a play.
The conductor has a monitor for the action, and a camera relays the
beat to the action studio, where a repetiteur takes it from another
monitor and passes it to the singers.
Sound communication is a little more complicated. Orchestral sound is
relayed to the action studio floor by a directional loudspeaker mounted
on the front of the boom dolly. As the boom is moved to follow action,
so is the upright line-source loudspeaker that is fixed to it. A line
source is directional at middle and high frequencies, radiating mainly
in the plane that intersects the line at right angles, so this allows good
separation from a high microphone. The voices of the singers can
be relayed to the conductor in a variety of ways: one is to have a
line-source loudspeaker along the top edge of the conductor's music
stand. Three small elliptical speakers set in a line produce little bass
and are again directional at high frequencies, helping separation.
Even using this technique, it may be necessary to pre-record or film
some scenes in advance. To employ the musicians for a short
sequence may be disproportionate, so a system of guide tracks
may be used. Here the pre-recorded (or filmed) scenes are accompanied by a soundtrack of bar counts or a piano version, plus any
verbal instructions such as warnings of impending change of tempo.
This guide is fed to the conductor only (by headphones in this case).
Orchestral sound is then added at the main recording.
If singers' voices are included in the pre-recording this can also be
used as guide track or, better still, separate tracks can be used. Where
singers' voices are pre-recorded in sound only (so that miming can be
used, perhaps in order to cover distant action), a guide track can
again be included in a multitrack recording.
The fourth arrangement, with orchestra and singers at opposing ends
of the same concert hall, employs many of the same communications
systems: for example, the acoustic time-lag between the ends of the
hall still requires the use of the conductor's music-stand loudspeakers.
Separate rehearsal, one of the advantages of twin studios, is lost, but
in its place is an enhanced unity of purpose.
Popular music
Having worked up to some of the largest and most complex layouts
for classical music, we can now take some of the same (or similar)
Music balance
151
components and put them together in a contrasting type of balance
that is used for rock, dance, jazz and much other popular music.
A simple pop group may consist of drums, guitars (lead, bass and
possibly acoustic), vocalist and perhaps a synthesizer. In the dead
acoustics of a big recording studio the various components might be
placed in booths around the walls, with tall sound-absorbent screens
on either side of the drum-kit (unless it is in a corner) and halfscreens in front. To reduce bass transmission, the drums may be
raised about 15 cm (6 in) on internally-braced low wooden boxes
(rostra, risers), with resonances further damped by laying them on
strips of carpet.
[8.46] Pop group
Three electric guitars, drums and
vocalist. For maximum separation the
drums may have two low screens in
front, high screens at the sides, and
a further screen bridged across the top.
For stereo, microphones are steered
to suitable positions on the audio
stage, and blended and spread
artificial reverberation is added.
Lead and bass guitar loudspeakers can be directed in to halfscreened bays, and picked up on cardioid microphones, but an
acoustic guitar may need a separate booth. The synthesizer is
angled to radiate any acoustic output into an opposite wall, and
the amount of screening depends on whether there are any additional, quieter instruments to be balanced: if so, further heavy
screening may be required. The guide vocalist is kept physically
well away from all of these but in good visual contact with the
other players.
Much use is made of tracking (sequential recording) for vocals,
vocal backing, additional acoustic guitar, percussion and other
effects. Some parts, such as the main drum coverage, keyboard
and vocal backing, may appear on stereo pairs of tracks; the rest
will be mono. One track may carry time code.
The session may start by recording the basic rhythm plus guide
vocal, all on separate tracks, followed by additional acoustic guitar
and the synthesizer. Next the lead vocals are laid down, then the
backing (as separate stages if complexity warrants this). Finally, the
solo guitar, electric guitar riffs, etc., are overdubbed. Alternatively,
all or most of the tracks may be recorded at the same time. The
mixer will have two faders for each channel, one of which may carry
the full audio signal, which is then given nominal (but not irreversible) treatment via the buttons at the top of the strip on its way to the
multitrack recorder, MDM (digital modules) or hard disk recording
system. The second fader, below the first, controls a separate feed
that is used to monitor a trial balance, and this may also be recorded
for later reference. In a fully digital console, the faders and all
controls mimic their audio counterparts, but the signal is not fed
directly through them.
Reduction (or mix-down), in a separate session, begins by returning
the tracks to the mixer channels, checking and refining their treatment, starting with rhythm as before, and gradually building up a
stereo composite from the separate components. A computer may be
used to record and update control data and make the final mix.
Multitrack recordings can also be used in this way for classical
music.
For radio, where less time and money may be available, the process
is often condensed, with the multitrack recordings used mainly for
152
Music balance
cosmetic changes, to correct errors or for the greater freedom to
experiment that is allowed by the possibility of second thoughts.
[8.47] Rhythm group
The drum-kit and bass are separated
by a low acoustic screen. A range of
positions is possible for microphones,
including several for the drum-kit.
To regain a sense of occasion and the feeling of live performance,
recordings may be made from concerts or road shows. A rock
concert can involve a great amount of electronic rebalancing, even
before any recording is contemplated. A group of three or four
players may have 30±100 channels, linked to a travelling mixer
(with a separate mixer for the drummer) to feed the stage speaker
system. Studio-style consoles send feeds to audience loudspeakers
and make recordings. These are taken back to the studio and
generally remixed in the normal way, though records produced from
concerts are sometimes proudly marked `no overdubbings'.
When recording on location, whether from a pop or classical
concert, time is limited and it is wise to establish a routine, setting
microphones and establishing any split feeds as early as possible.
Microphones are identified by scratching them and indicating what
instruments they cover. Radio microphones may be used for direct
communication between stage and mixer. The recording arrangements are set up and checked out. A mobile sound control room is
often used.
Larger groups
A basic component of much light music, including jazz, is the piano
trio, i.e. piano, bass and drums. It is also the nucleus in many larger
groups. To it are added, stage by stage, guitar (acoustic/electric),
brass, woodwind (including saxophones), singers and other instruments for particular effects. These might include celesta, violin,
flute, horn, harp, electric organ or any other instrument that interests
the arranger. As usual, mono microphones are used for most instruments, but stereo pairs might be tried on trumpets, trombones,
woodwind, xylophone, piano and vocal groups. Again, tracking is
generally used.
[8.48] Small band
Rhythm group with: 1, Guitar
(acoustic/electric). 2, Trumpet.
3, Trombone. 4, Woodwind
(saxophones). The rhythm group
microphones are not shown.
Often the studio is dead, but sometimes only bright surroundings are
available. If some spill is inevitable, it might be best to modify the
ideal fully separated balance. Even in good conditions there may be
enough spill in the studio to dictate some elements of the stereo
sound picture. However, certain relationships will always appear in
both the real layout and its image: for example, the piano, bass and
drums are always close together. These might be steered to centre
and left of centre with electric/acoustic guitar near the left speaker,
singer and woodwind right centre, and brass on the right. The
actual layout in the studio is governed not only by the need for
separation (so that this or some other stereo picture can be achieved,
and so that all sounds can be individually treated), but also by
practical considerations, such as whether the players can hear or see
each other.
Music balance
153
Multimicrophone layout
When musicians arrive in the studio they expect to sit at positions
already arranged for them, with chairs, music stands, screens, piano,
boxes for such things as guitar amplifiers and microphones all (at
least roughly) laid out. They will have been set according to the
balancer's instructions, and the balancer or an assistant will have
placed the microphones. Three different situations may apply:
.
.
.
[8.49] Show band with strings
1, Rhythm group. 2, Electric/acoustic
guitar. 3, Trumpets. 4, Trombones.
5, Woodwind. 6, 1st violins. 7, 2nd
violins. 8, Violas. 9, Cellos. 10, Vocalist.
11, Musical director. Variations must be
made for acoustic difficulties: in one
layout it was deemed necessary to bring
the bass forward to the tail of the piano
and separate it by acoustic screens
from brass, drums and woodwind (the
trumpets having been moved farther
away). Additional forces complicate the
layout further. For example, the pianist
may need to reach a second keyboard,
for organ or celesta: this could be
placed to the player's right and the
guitar moved forward.
The group may already have a layout used for stage shows. The
balancer finds out in advance what this is and varies it as little as
possible (but as much as may be necessary for good separation)
with the co-operation of the musical director.
The group may have made previous recordings or broadcasts in the
same studios. The balancer finds out what layout was used before,
and if it worked well, uses it as a basis for this one: the group will
not expect to be in different positions every time they come to the
studio. But there may be changes for the purpose of experiment, or
to include ideas suggested by a producer or which the balancer is
convinced work better than those used previously: these may
require the co-operation of an individual player, section leader or
musical director. Bigger changes may be dictated by the use of
additional instruments or developments in musical style.
The group may be created for this session only, in which case the
balancer ± having found out the composition from the producer
or musical director ± can start from scratch.
154
Music balance
Even starting fresh, the balancer can apply certain rules that make
the musician's job easier. For example:
.
.
.
.
.
.
.
.
[8.50] Studio screens
A dead music studio has a supply of
full and low screens with absorbers
about 10 cm (4 in) thick that are easily
moved on rollers to provide acoustic
separation without loss of visual
contact. The vocalist's screen may
have a window.
The bass player may wish to see the pianist's left hand, to which the
bassist's music directly relates. This is important in improvisations.
The drummer, bass player and pianist form a basic group (the
rhythm group) and expect to be together.
In a piano quartet the fourth player can face the other three in the
well of the piano; everybody can see everybody else. But a
guitarist may work more with the bass, so could be placed at
the top end of the piano keyboard, to see across the pianist to the
bass player.
A bass player with fast rhythmic figures needs to hear the drums,
particularly snare drum and hi-hat; these can usually be heard
well on the drummer's left.
Any other two players who are likely to work together on
a melodic line or other figures need to be close together.
When players double, e.g. piano with celesta, the two instruments
must be together.
All players need to see the musical director or conductor.
The musical director may also be playing an instrument.
This may appear to make the original idea of building up a layout
according to good principles of separation less feasible. But
although compromises are often necessary, the use of directional
microphones and screens should make a good sound possible. Low
screens are set between bass and drums, and more to separate drums
and brass. Other sections, especially quiet instruments such as
strings or celesta, will also require them. A singer may be provided
with a tent of screens well away from the players, but should still be
able to see them.
Popular music in vision
Plainly the typical recording studio layout for multitrack music
cannot be adopted in vision (though a band wholly or largely out
of vision will). But apart from visual layout ± which may produce
some problems of separation ± all other elements of the balance and
treatment will be as close as possible to that of the recording studio,
which inevitably will be the standard by which the television sound
will be judged.
Low microphones in among the instruments are not obtrusive: they
are readily accepted as part of pop paraphernalia. Large microphone
booms and stands are avoided, as are microphones at or just above
head height. There are more differences in mountings than in
positioning.
The pop singer's hand microphone is accepted by the audience as a
prop and not just as something to pick up sound, though some
experienced performers also use it to balance themselves with skill.
Music balance
155
[8.51] Show band in vision to
support singer or featured acts
One of many possible layouts:
V, Violins. Va, Violas. C, Cellos.
DB, Double bass. G, Acoustic
guitar. T, Timpani. P, Percussion.
D, Drum-kit. H, Horns.
Tp, Trumpets. Tn, Trombones.
Ac, Accordion. S, Saxophones,
doubling woodwind. W, Overall
wind microphone. All
microphones may be condenser
cardioids except for a moving
coil on kick drum.
For the less mobile singer, a stand microphone is an accepted
convention.
For any popular music television programme, the control console is
likely to have most of the equalization, compression and AR facilities of the recording studio. If several groups are to appear, a
separate studio desk may be used to balance each in turn. Digital
`snapshots' of the rehearsed mixer settings can then be recalled as
each group appears in a live show. It may be necessary to pre-record
some tracks even in a live performance if a group is to match its
expected sound. Miming to playback of records is cheap and easy ±
but some would say cheap and nasty: the sense of occasion is
prejudiced, and the audience may feel cheated. Such devices should
be kept to a minimum.
Television has largely adapted its pictures to the sound requirements
of popular music, but sometimes a singer must appear without a
microphone in vision. Booms may still be used for this, perhaps with
a gun microphone to improve separation. Only where separation
cannot be achieved by these means, or in complex song and dance
acts, should the music be pre-recorded and vision shot to playback of
sound.
Much of this applies also to music on film, although there may be a
greater effort to eliminate microphones from the picture.
Although multiple-camera techniques have, at times, been used for
music on film and location video, close-ups are usually shot separately, taking sound as a guide to synchronization with a master
recording. More is shot than is actually required, so actions before
and after cuts will match. Sometimes film shot in sync for one
sequence is later synchronized to the sound from another part of
the musical work. For this, the beat or action, e.g. drumbeat, is laid
down. It can then be matched to a close-up of the action.
Chapter 9
Monitoring and
control
Monitoring means using ears, and therefore loudspeakers (or, at a pinch,
headphones) to make judgements on sound; control is the use of hands
to manipulate the sound, mainly in a mixer console. This chapter
describes loudspeaker layouts; the path of the analogue or digital signal
through typical consoles; and touches on questions of quality, without
which the more creative modification of sound is wasted.
During rehearsal, recording or broadcast, the ear should be constantly searching out faults that will mar a production. For example:
.
.
.
.
production faults ± e.g. miscast voices, poorly spoken lines,
uncolloquial scripts, bad timing;
faulty technique ± poor balance and control, untidy fades and
mixes, misuse of acoustics;
poor sound quality ± e.g. distortion, resonances, and other irregularities in frequency response, lack of bass and top;
faults of equipment or recordings ± wow and flutter, hum, noises
due to loose connections, poor screening, or mismatching.
Of these, the faults in the first group are the most important. They
are also, perhaps, the easiest to spot. As for techniques, the ability to
see faults in those comes with their practice. But the various things
that can be wrong with the quality of a sound ± supposedly objective,
measurable things ± appear, surprisingly, to be the most difficult for
the ear to judge in other than subjective `I-know-what-I-like' terms.
Quality and the ear
The quality of sound that the ear will accept ± and prefer ± depends
on what the listener is used to. Few whose ears are untrained can
Monitoring and control
157
judge quality objectively. Given a choice between loudspeaker
qualities, listeners will generally state a preference, but in practice
it seems that as many prefer medium quality as want the best. For
many, the preference is for a box that sounds like a box: the kind of
quality that turns their living-space into a corner of a great concert
hall is disturbing. But while such preferences must be respected, it
must also be true that quality and enjoyment are often closely
related. Higher quality replay of a musical work gives more information, so it is often a technically superior rendition that first `sells'
unfamiliar music. And a good loudspeaker allows more of the
original intention, as expressed both in content and performance,
to reach the listener.
Different levels of quality may be acceptable for different purposes.
High quality may be best for brilliant modern music (or for orchestral music heard as a concert); medium quality can be satisfactory
for ordinary speech, where the main objective is to hear the words
clearly; and `mellow' quality is often felt to be less obtrusive in
music intended as a companionable background noise. On highquality equipment these different types of programming are all
available at the appropriate levels of quality; on poor equipment
they are all debased to the same indifferent standard. It is undoubtedly worth persevering with the education of one's ears, and also
allowing others the opportunity of listening to sound that is of better
quality than they are used to.
Nevertheless, any organization that is conscious of cost must decide
on a practicable economic upper limit to the audio-frequency range
to be offered. In a test by a broadcaster, the range was restricted
successively to 7, 10 and 12 kHz. Sounds in the test included tubular
bells, brass, cymbals and triangle, snare and military drums, maracas, handclaps, and female speech. The listeners were all less than
40 years of age, both male and female, and included some who were
experienced in judging sound quality. The results suggested that:
.
.
.
only a few critical observers hear restriction to 12 kHz;
many inexperienced listeners can detect restriction to 10 kHz,
while experienced listeners do so readily;
a filter at 7 kHz was detected even by the inexperienced.
In fact, although aiming to record a full range from 20 Hz to 20 kHz,
that broadcaster actually transmits only 30 Hz to 15 kHz on FM radio
and NICAM television stereo. For CD and digital tapes the recording
industry goes a step further, up to 20 kHz.
[9.1] Loudspeaker types
Above: Moving coil. The most
commonly used for all purposes
(bass, treble and wide range).
1, Permanent magnet. 2, Moving coil.
3, Cone. 4, Baffle or enclosure.
Centre: Ribbon unit. Sometimes used
for treble. 5, Permanent magnet.
6, Ribbon (seen in plan view).
7, Acoustic horn. Below: Electrostatic.
8, Perforated fixed electrodes.
9, Diaphragm.
Loudspeakers
For technical monitoring, balancers need high-quality sound ± which
is actually quite difficult to achieve. The main obstacle to good
sound has nearly always been at the final stage, where an electrical
signal is converted back into waves in air. It is not so difficult to
design for a good middle- and high-frequency loudspeaker response:
158
Monitoring and control
the hard bit is to combine that with efficient production of the bass.
Long-established designs of `woofer' use either a fairly large surface
area to push the air, or (even longer established but now little used
except in traditional musical instruments) a horn that gradually
flares out to a large aperture. (Separate subwoofers extend the range,
but this may be more for effect than to increase quality.)
Most loudspeakers use a stiff diaphragm working as a piston. It is
operated by an electromechanical device that is similar to those used
in microphones ± except, of course, that it converts an electrical
signal back into physical movement. Although many other principles are feasible, the most common type is moving coil.
[9.2] Loudspeaker: simple
crossover network
1, Bass unit. 2, Treble unit. 3, The total
power is divided evenly between the
two units 4 and 5. For each it falls off
at 6 dB/octave beyond the crossover
point. This is a very satisfactory type of
network provided that there are no
severe irregularities in the response of
either unit that are close enough to the
crossover point to make a significant
contribution even though attenuated.
For example, a severe fluctuation in a
bass cone output at A will still affect
the overall response. To counter this,
the crossover point might be taken
lower or the dividing network might be
redesigned to give a steeper cut-off at
the crossover.
[9.3] Loudspeaker cone with sides
of corrugated paper felt
The performance diagram shows
axial frequency response with output
in decibels (arbitrary scale), taken
from a particular example.
It is not easy for a single diaphragm to respond equally well to all
frequencies: usually the signal is split and fed to separate speakers
covering different parts of the frequency range. When the sound is
divided like this there is no need for a sharp cut-off as one loudspeaker takes over from another, so long as the acoustic output adds
up to a full signal. Careful relative positioning of the units is
required, to avoid phase cancellation at switchover frequencies.
(But the subwoofer in surround sound seems exempt from this rule:
its separate existence isolates it from awkward crossover effects.)
For the low-frequency unit the diaphragm may be a stiff cone of
paper, corrugated concentrically in order to stop unwanted harmonics from being formed (like a bell, these may include a subharmonic, below the frequency at which the cone is driven). Pistons of
other materials are used: again they must be stiff, but with internal
damping or they may ring like a wineglass. Of whatever construction, a driver must also be of low mass or its inertia may affect the
response.
A feature of loudspeaker design is that as the cone or diaphragm
radiates the wanted sound from the front, a second component,
opposite in phase, is emitted from the back, and if there is nothing
to stop it, some of this back-radiation leaks round the sides and joins
that from the front. Long-wavelength low frequencies are most
affected by this: as front and rear combine, they are still partly out
of phase, so tend to cancel sound that is already weaker than it
should be because any cone is too small for effective low-frequency
work.
In the design of high-quality loudspeakers there have been many
ingenious part-solutions to the problem posed by the size naturally
associated with the longer wavelengths of sound. The simplest is to
mount the loudspeaker in a much larger baffle (a fixed solid panel).
A baffle 75 cm (30 in) square is about the smallest for a modest lowfrequency response; a whole wall is better. Another idea is to turn
the space behind the cone into a low-frequency resonator with a port
that re-radiates some of the power (in this vented box the neck acts
as part of a tuned Helmholtz resonator). By changing its phase this
can be made to boost the response in the range where it previously
fell off. But this `bass reflex' loudspeaker is bulky, may be subject to
cabinet coloration in the bass, and below its resonant frequency cuts
off even more sharply than before. It also has a characteristically
Monitoring and control
[9.5] Bass reflex loudspeaker
This uses the internal cavity and a
port at the front as a Helmholtz
resonator with a low-frequency
resonance peak. Added to the normal
bass response of the loudspeaker (1),
it can be used to lift the bass for about
one octave (2). Below this (3), the
cut-off is more severe.
[9.4] Loudspeaker housings
Above: Baffle. Below: Enclosure
with an opening at the rear. Both
reduce the frequency at which
sound radiated from the back of a
cone can recombine with and cancel
the direct sound. A box may have
internal resonances that need to be
damped by lining it with absorbent
material.
159
[9.6] Infinite baffle enclosure
The sound from the rear of the speaker
is trapped and much of it is absorbed.
In practice, the stiffness of the air
within the box somewhat restricts the
movement of the speaker. The cavity
contains folded sound-absorbent
material.
recognizable quality ± not a recommendation for any loudspeaker.
Yet another idea is to lose the back-radiation completely by stifling
it within the box so the diaphragm is not affected by reflections or by
the `stiffness' of the enclosed air.
Powerful loudspeakers require additional separate amplifiers to
drive them. Without these, in the bass particularly, overmodulated
sound on which the signal is clipped will create strong harmonic
distortion products that are heard on higher-frequency loudspeaker
elements.
Monitoring layout
The first question in control-room layout is architectural: whether or
not to let listening room acoustics dominate its design. In the more
extreme examples (most often found in the music recording industry) the loudspeakers are no longer free-standing boxes (ideally, well
away ± e.g. a metre ± from any wall), but instead are recessed in
cavities within the walls. The room is not a rectangular shape, but
instead has walls and ceiling that flare out symmetrically and round
to the sides, with absorbers and diffusers to break up potential
coloration. First reflections continue to disperse, and some are
soaked up around the back. And producers (who are paying for this)
160
Monitoring and control
are discouraged from taking what often seems to be their preferred
position in the control room ± which is tucked away in a remote
corner with the telephones.
Many control rooms are more utilitarian, and do have free-standing
loudspeakers. It may be best for the balancer and producer not to see
directly into the studio as they face the sound stage, but instead sit
where they can turn to see and sort out positioning or other problems. Adapted rather than purpose-built, the room may have irregular shapes and intrusions (often including equipment racks) that
disturb a balanced sound field. It will usually have some acoustic
treatment, and will certainly be isolated from any studio with microphones. Noise, both coming in and heading out, will be controlled.
(And visiting producers and directors will still need somewhere to
talk on the telephones.)
[9.7] Complex loudspeaker
enclosure
The tweeter (HF unit) has an infinite
baffle and the woofer (LF unit) backs
on to an acoustic labyrinth (also called
a transmission line or pipe). This
introduces both phase change and
attenuation. Again, both cavities are
padded.
A minimum distance from which to listen to a loudspeaker is 1±2 m
(3±7 ft). Closer than this, the sound has an unreal quality owing to
standing waves near the cabinets; at a greater distance the acoustics
of the listening room begin to take over (so larger rooms are made
more dead). Except for film dubbing theatres, which are designed for
a much bigger (and usually surround) sound, it is desirable that these
acoustics should be similar to those of a domestic living room. But it
is important to produce both a well-blended signal and to be clear on
whether some minor fault is due to the studio or listening room, and
the only way to be sure is to sit no farther from the loudspeakers than
when listening for pleasure at home.
For monitoring two-speaker stereo the balancer should be at the apex
of an equilateral triangle, or along the centre line from a little closer
than that to not more than 1.7 times farther back. If the loudspeakers
are about 2.5 m (8 ft) apart the balancer will therefore be 1.8 m (6 ft)
or more from them. For 5.1 surround sound music balance the layout
of the five main speakers is as shown in Chapter 3: these five should
all be of the same type and set-up, but the subwoofer should be set
for 10 dB higher volume (which is quite hard to judge by ear). For
film, the side speakers may be `bipolar', radiating fore and aft for a
more diffuse effect, and set not so far back, at 90 to the balancer.
The producer or director should ideally sit just behind the balancer
(but may choose not to). Because monitoring loudspeaker volume is
sometimes set rather high, there is an attenuator of, say, 20 dB that
can be cut in (by a `dim switch') when using telephones, for example.
During talkback to the studio it generally switches in automatically.
[9.8] Monitoring stereo
A, High-quality loudspeakers.
B, Balancer. C, Producer, also
equidistant from the loudspeakers.
D, `Mini-speakers' offering a
distinctively different lower quality.
For stereo an additional pair of small loudspeakers is sometimes
mounted closer and in line with the monitoring loudspeakers. These
`mini-speakers' (or near-field loudspeakers), set at lower volume,
serve to check what many of the audience will actually hear: all vital
elements of the high-quality balance must still be audible when
output is switched to them. They are also used for talkback and for
listening to channels that have not been faded up (pre-hear or prefade listen, PFL). After-fade listen (AFL) takes its feed later, as the
name implies. Switching over to these small speakers does not affect
output. On some consoles the PFL for any channel can be switched
Monitoring and control
161
in by pressing on a fader knob when it is fully faded out: this is
called overpressing.
[9.9] Pre-fade listen (PFL, pre-hear)
1, Output is taken before channel
fader. 2, Symbol for `overpress'
facility: when channel is faded out,
pressing the fader knob routes the
signal to the pre-fade listen condition.
Audio can then be heard on near-field
or studio loudspeakers as selected, or
on headphones. Other options may
be available: for simplicity, these are
not shown on the circuit diagrams.
Most consoles have a wide range of further switches associated with
the loudspeakers. `Solo' selects just the channel in which the switch
is located and feeds it to a main loudspeaker; `cut' takes out just that
one channel, to hear what the mix is like without it. On some
consoles it is possible to solo one channel while retaining another
channel or group for background or comparison: for this, press `solo
isolate' on the second fader at the same time. Some consoles also
allow `solo in place' ± that is, in its stereo position. If these operate
before the fader, the individual balance can be checked; if after, that
fader's contribution to the overall balance is also assessed. If operated during rehearsal, these switches do affect studio output, so they
are linked to the transmission or recording condition that is set up in
the console when the red light is on: solos, etc., will then still be
heard, but on the small loudspeakers, as for PFL. There will also be a
headphone feed.
Control consoles and racks
Sound control consoles (otherwise called mixer desks or, sometimes,
panels) range in complexity from the simplest that are used in radio
for speech and record shows to those for popular music recording,
which may have a vast number of channels, each with the facility for
a range of treatments and two separate mixing systems so that one
can be used to monitor (or broadcast) the music being played while
the other feeds a multitrack recording system, for subsequent remixing.
[9.10] Mono channel
Some of the items that may be found
before the channel fader:
1, Microphone (low level) input.
2, Phantom power feed to microphone
(48 V). 3, Isolating transformer.
4, Microphone channel amplifier and
coarse pre-set. 5, Line (high-level)
input: in this example, replay of
recordings. 6, Buffer. 7, Fine pre-set
gain (typically ‡15 dB). 8, Polarity
reversal control (may be called
`phase'). 9, Equalizer. 10, Equalizer
bypass switch. 11, Optional `insert'
routing: allows external module to
modify channel signal. 12, Pre-fade
listen routing. 13, Remote cut switch,
or `cough key'. 14, Channel fader.
The size (or complexity) of a console reflects the maximum
demands expected of the studio. In a large broadcasting centre, radio
studios may be more specialized than those for television. A television studio represents a large capital outlay and is used for a greater
variety of productions, and the demands of vision add to the complexity of sound coverage, so television consoles may need to be
more versatile than their radio counterparts. Consoles used by the
record industry may be more complex still. But this size can be
deceptive: however big they are, most consoles are designed for use
by a single pair of hands. Much of the space on a big desk is
occupied by controls that will be pre-set; other adjustments are made
one or two at a time, and continuous control will be exercised only
over a narrow range near the centre of the console. This, at least, is
162
Monitoring and control
[9.11] Stereo channel
Some facilities before channel fader:
1, Stereo microphone. 2, Buffers.
3, Coarse channel amplifiers. 4, Channel
input switching (to mono). 5, Polarity
reverse switching in left leg only.
6, Equalizer. 7, Equalizer bypass.
8, Optional `insert' for plugged modules.
9, Pre-fade listen. 10, Remote cut.
11, Ganged channel faders.
the ideal ± and as balancers devise increasingly complex sequences
of operations, console designers have come up with ways of building
these up (or changing them in detail) stage by stage, and of also
memorizing and recalling earlier settings.
At one time all consoles carried only analogue sound, and the size of
the biggest reflected the great number of individual inputs that must
be separately controlled. As the number of things that could be done
to each input (or channel) increased, the clearest layout had all the
buttons, knobs, levers, switches, meters and flashing lights for each
channel arranged in an apparently logical strip up the desk (although
many of the components they controlled, such as amplifiers, were
somewhere else in the desk, or in racks). Even at the first stage of
control, with so many of these channels side by side, this already
covered a large area on the desk.
The key to understanding how a big console works is, first, to work
out what happens in a typical individual channel; second, how
channels fit into a hierarchy so that a much smaller number of later
channels each allows control over a group of those earlier in the
chain, followed by further grouping, to end up with a master output
control. Typically, group and master controls should be near the
middle of the desk, with larger banks of the earlier individual
channels on either side.
A third, smaller group of controls must also be (literally) close at
hand. These include: switches, etc., for additional operational
devices that may be inserted into selected channels; routing switches
and controls for audio feeds to be sent outside the console or beyond
its central hierarchy of channels; monitoring controls; and communications. Most of these are in a block that should also be easy to
reach, but which can somewhat untidily break up the neat repetitive
array of channels.
Racks in the control room or a linked machine room accommodate
any equipment that does not fit on the console or which requires easy
access or ventilation. (Noisy, fan-cooled or mechanically operated
items of equipment need a separate machine room.) Racks may
house amplifiers, devices for treating sound, play-in equipment
and recording decks, and also cross-plugging bays ± patch bays.
A `rack' is a collection of items of equipment made to fit into
a frame with supports that are 20 in (508 mm) apart. Each unit has
Monitoring and control
163
flanges at the front, 19 in (482 mm) wide, pierced by holes used to
screw it onto the frame. Patch bays are slotted into this at a comfortable access height.
[9.12] Patch bay (jackfield)
A standard-sized unit with two rows,
each of 24 jacks. There may be
several of these, mounted directly in
the console, or a whole telephone
exchange of them mounted in an
equipment bay. Jacks are wired in to
intercept signal paths at a variety of
points in the desk. Often (but not
always) the upper and lower jacks are
`normally' connected together inside
the panel ± `normalled', that is, when
nothing is plugged in to them. These
pairs act as an `insert point' to a
channel (often just after the equalizer)
to allow an external device to be
plugged in. In a `half-normalled' pair,
the upper jack can be used to plug in
headphones without breaking the
signal path, but the pair still works as
an insert point, diverting its path, if
there are plugs in both jacks.
A patch bay is made up from a large number of units, each with 20 or
24 pairs of jacks (or 48 smaller jacks) which, ideally, are connected
to as many points as possible throughout the desk. Inside some of the
units, the upper and lower jacks in each row are independent
(`open'): these serve as inputs to each of the channel lines on the
desk. Microphones, replay machines and lines from outside the
studio can be plugged in to these. In other units the upper and lower
jacks are `normalled', which means that unless plugs are inserted,
the jacks are linked together. Since the two jacks are connected to
two adjacent points in the control console, this allows an external
device to be plugged in between them. The top rows are `sends', the
lower rows `returns'. A `half-normalled' variation on this allows the
top row (only) to be intercepted without interrupting the internal
connection, often so that the circuit can be monitored at this point.
Another variation, called `parallel', allows both rows to be intercepted without breaking the connection.
Other racks contain `sends' and `returns' for all the individual tracks
of multichannel recording devices, plus `in' and `out' jacks for
external devices and for other recorders, and jacks to intercept group
and main controls, etc. The bay acts like a highly flexible telephone
exchange ± one that actually uses traditional telephone plugs
and cables.
Microphones are often hard-wired from the studio to racks where it
is possible to intercept and cross-plug them. For consoles that
require a digital input, the analogue signal may pass though a
separate analogue-to-digital (A/D) converter or one that is included
in the console.
Digital conversion
Up to this stage in treating sound, analogue electrical signals, in
which the waveform mimics that of sound in air, have been unavoidable, if only because microphones and loudspeakers work directly
with those waveforms. But as the signal is fed to the console there is
an option. While it is possible to continue to work with analogue
sound (many excellent consoles are designed to do just that) the
alternative is to convert to digital data and operate on that instead.
Here we need to make a short diversion ± to see what `digital sound'
is, and then how the analogue version is converted to digital. We
might also ask `why?' There would be no point in introducing two
extra stages (including reconversion, taking it back to analogue so
that we can hear it) unless there are purposes for which it is better.
First, what is `digital' audio? Although it has been described as `pure
information', that is simplistic. A sound wave in air also carries
information: it is what microphones convert into electrical `analogue'
164
Monitoring and control
audio. At one time it was argued that computers might reflect what
happens in Nature better if they worked directly with analogue forms
of information. But measurement uses numbers, numbers are digital
and, as it turned out, digital computation soon became faster, so it
was possible to do more, at ever increasing rates, and mostly in `real'
time ± a big advantage for audio. Ever-increasing amounts of
information could be laid down in compact forms, and retrieved
with lower losses than with analogue forms of storage. And information could be transmitted more accurately, with enough redundancy
of information that damaged data could often be reconstructed.
Next, how is `analogue' converted to `digital'? The `information' in
both cases is about how a waveform changes in time. In `analogue'
there is simply a `picture' of the wave. Digitization means converting
that shape into a series of digits (the numbers 0 and 1) by sampling the
height of the wave often enough to define its shape, even if it makes
20 000 excursions up and down in a second. There is a mathematical
rule about how often we need to sample a waveform in order to be
able to reconstruct it perfectly: according to the Nyquist theorem, we
must sample it at least twice as often as the highest frequency in it.
This is essentially what an analogue-to-digital (A/D) converter does.
[9.13] Digitization
1, Analogue waveform. 2, Digital
sampling: for DAT (a digital audio tape
recording) this takes place 48 000
times per second. 3, When converted
back to analogue form, the data
traces out an irregular shape that
approximates to the original but with
higher harmonics that will be
eliminated.
A common A/D sampling rate is based on the supposed 20 kHz
range of human hearing. The actual sampling rate used for this is
not 40 kHz, but 44.1 kHz ± the extra margin allows for the sampling
to operate at full strength over the full audio range, then tail off in
the extra space allowed; any filter should also operate in this marginal range. A 44.1 kHz sampling rate has been adopted by the
record industry for CDs.
Digital audio tape (DAT) generally has a more generous 48 kHz
sampling rate ± on the principle that if professional sound reproduction goes beyond any reasonable limit to human hearing, the
response at lower frequencies that people can actually hear is more
likely to be smooth. However, in studios where output may be
recorded on CD-R, it is simpler to use a common 44.1 kHz for both.
Yet another rate, 32 kHz, may be encountered where 15 kHz is
accepted as a more practicable limit to transmission of VHF/FM
audio and NICAM stereo: the argument here is based on bandwidths
(more transmission channels are available than with a wider bandwidth) and on cost. A sound card on a computer might offer all these
and other, much higher A/D sampling options.
Many recent recording systems offer a range of sampling rates. For
example, a journalist's hand-held recorder can be set for 16, 24, 32
or 48 kHz (mono or stereo). Here the solid-state recording medium
is a random access memory Flash card that fits into a PCMCIA
slot inside the recorder. A two-channel portable recorder with a
removable 2.2 GB hard disk offers a choice of 44.1 or 48 kHz.
A four-track digital DAT recorder can sample at 32, 44.1, 48, 88.2
or 96 kHz. Yet another format is DVD-RAM, sampling at up to
192 kHz for four-channel stereo.
Digitization is technically complex, involving many further stages
and processes. For CD quality, the height of each sampled point in
Monitoring and control
AES31 is intended as an
industry-wide set of audio
interchange standards. The part
most likely to be directly encountered by users is AES31-3. It is
designed to use EBU 'Broadcast
Wave' files (i.e. Wave audio files
with added descriptive data) and
contains just about everything
needed for simple audio project
transfer
between
working
studios with deadlines to meet.
165
the waveform is described by a series of digits (0 or 1): 16 of these
make up a 16-bit `word'. But to calculate with this and finish with an
equally accurate set of 16-bit words at the end of the process requires
intermediate stages with at least 18- or possibly 20-bit words. But
more bits also means a greater signal-to-noise ratio, which allows
increased dynamic range or a bigger safety margin (`headroom')
before destructive clipping occurs. So at this intermediate stage it is
better to work with 24-bit words. There are several processes that are
used to minimize, check for and try to correct errors. For CD or
broadcast, the end-product of all this computation ± with both
dynamic range and headroom under tight control ± must eventually
be scaled back to 16 bits. Paradoxically, the outcome of word-length
reduction is improved by introducing noise (`dither'), so that some
of the numbers are randomly rounded up and others rounded down.
When converted back to analogue, this gives a more realistic
waveform.
For sequences of these numbers to be multiplied together they must
come together at just the right time, which is not a problem if they all
originate from one source, but can be if they come from several (say,
a bank of recorders). With multiple sources, one sets the time and its
`clock' is sent (often by a separate cable) to all the others.
To send a digital signal along wires between components or to
greater distances by cable has required the definition of a variety
of standards and formats (plus a further vast swathe of the inevitable
acronyms). There is AES/EBU, also known as AES3: on a suitable
cable with XLR connectors this will carry two channels 60 m (200 ft)
± but its word clock wire should be no more than 6 m (20 ft). The
semi-professional S/PDIF is for two channels with word clock; TDIF
is a commercially originated standard for carrying eight channels
each way locally (useful for modular digital multitracks, MDM); and
ADAT is good for eight each way by optical cable (`lightpipe').
Wave and AIFF are two professional formats for uncompressed hard
disk recordings (typically 16-bit, 44.1 kHz, PCM format). The computer file suffixes are `.wav' and `.aif'.
[9.14] Journalist's solid-state digital
recorder
Digital equipment may be technically
complex, but this device is designed
for operation by a user who is
thinking of something else. The
`Flash card' slots in inside the case
and is forgotten until a warning
appears. Volume control is by `low',
`medium' and `high' signals that
interpret a simple bar meter. The
microphone may be on a short cable
or its capsule plugged directly into an
XLR connector on the recorder. The
main operational question becomes
how far it should be held from the
speaker's mouth (see `interviewing',
Chapter 7). The Flash card can be
transferred to a PC that has a
PCMCIA slot and can be edited
non-destructively.
To reconstruct an audible analogue sound requires digital-to-analogue (D/A) conversion. It reassembles a waveform, initially in the
form of a series of stepped blocks. This contains all the frequencies
required to define the required waveform, but its jagged, blocky
shape adds in a lot of extra harmonics at higher frequencies that must
be filtered out.
Digital signal processing (DSP) is superior to the alternatives for
both artificial reverberation and delay-related effects (see Chapter
11), and provides a competitive alternative for many other functions
in and around the console. It is superior for broadcast and transmission by landline. Also (apart from some `retro' preferences), it has
taken over almost the entire field of recording (see the brief reference later in this chapter). And most professional editing now has the
benefit of digital computing power and memory (Chapters 18 and 19).
We may add a fourth question: what can go wrong? One answer to
that lies in a cautionary tale from the early days of digitization, when
166
Monitoring and control
for some equipment that was distributed, the gap between the 40 kHz
assumed maximum and the 44.1 kHz sampling rate turned out to be
too small. If frequencies in this gap are not filtered cleanly at the
Nyquist frequency (half the sampling rate), they are in effect folded
back into the top of the audio band. This creates a spurious signal,
unrelated to the harmonic structure of the sound, a form of distortion
called `aliasing'. When this fault appeared it was eliminated by
replacing the filter with one that had a better cut-off. But the
manufacturers had to offer this free of charge for users to retrofit.
In such a rapidly advancing field as digital technology, users may
not have to wait very long for the next problem to come along. Bugs
in new software are corrected eventually in software updates, but
only after a period of frustration by users. Fortunately for the users
of analogue equipment, the technology time-lag is longer. Good
analogue hardware may be expensive, but it generally has a longer
useful life than either the hardware or the software in the everchanging world of digits.
Automated and digital consoles
In many consoles, whether the audio signal is digital or not, many of
the more complex control functions can be exercised by computer. It
may be used to set the console to some preferred initial state, or to
one of many intermediate `snapshot' states that have been worked
through. In the most extreme case, the entire provisional mix can be
reproduced, as though by many hands working with inhuman speed
and precision, but still allowing manual override (or just minor
`trimming' changes) to create yet another generation of operational
control.
A fully digital console goes one (big) stage further: all the audio
signals fed to it are converted into digital data, and all the controls
are digital too. A fully digital console can be far more compact than
an analogue console of comparable complexity, because sets of
controls can be `layered', set to switch from one set of channels to
another, or can be switched to different kinds of operations. Some
(designed primarily for the flexibility required in a television studio)
have a button switch in each channel fader, to switch the whole
channel to operate on a different source, and above the fader four
`wild' controls that can be switched to bring any of a wide range of
processors (set on another part of the desk) to operate on the signal
in that channel. Digital consoles vary more in layout and operation
than analogue consoles, so it may take longer to learn to use a new
one quickly and efficiently. But because the vast array of high-quality
analogue hardware can be expensive, digital generally has the edge
on cost as well as being more flexible. Even so, many users of digitally
controlled analogue desks still prefer them to full digital operation.
A radical alternative to this is to dispense with the console completely and allow a computer, with mouse, keyboard and one or more
screens, to do the lot. This is also what lies at the heart of the `virtual
Monitoring and control
167
studio'. In fact, what the virtual-studio computer does is actually
very similar to what all the consoles described here are doing: it is
just a `black box' within which the same operations are concealed,
then once again revealed as screen operations. In fact, the screens are
often visually designed to mimic the manual operations they replace
(and as a move in the opposite direction, there is a tendency to
reintroduce the hands-on equipment for some operations and plug
it in to the computer). But if there is a difference between real and
virtual studios it is little to do what the controls do, more a matter of
what the studios are used for.
Virtual studios mainly grew up in the world of synthesized music,
and in this context are discussed later, in Chapter 16; the use of
computers for editing sound is discussed in Chapters 18 and soundwith-video in Chapter 19. But many of the actual functions, which
are the same whether analogue or digital, are described here and in
the next few chapters.
Channels
The first thing to note about the channels on an analogue console is
that they are generally arranged so that they can be worked together
in pairs for two-channel stereo. The adjacent channels may be linked
(or `ganged') by a clip attached to the faders, so they are never out of
step by more than the fader manufacturer's `tolerance': a difference
that should never be greater than 2 dB. However, at this stage of
what may eventually be a stereo or surround sound mix, there will
also be many channels that are still operated individually, as mono.
[9.15] Analogue and digital console
channels
A, Analogue: part of the strip
dedicated to a single channel (there
may be two faders in line).
1, Monitoring fader and associated
controls. 2, Auxiliary sends.
3, Equalization and filters. 4, Dynamics
module. 5, Channel assignment.
D, Digital: each whole strip `flips'
between two `layers'. 6, Fader.
7, Layer selection with window
showing channel name above, backlit
amber or green to show layer
selected. 8, Four `wild' rotary
controls, each with a display above to
show its function or setting. Auxiliary
send, EQ and filter, dynamics, etc.
are assigned to these from modules
on other parts of the console.
There are two distinct kinds of input. One is from the microphone:
this will start at the point where the microphone feed arrives at the
input, often hard-wired from a three-pin female XLR socket in the
studio. There will often be provision in each microphone channel on
the console for a 48 V d.c. supply to feed back to a condenser
microphone, using the same three-conductor system that carries the
signal (a phantom power supply). The incoming signal passes
through a preamplifier that has a coarse rotary gain control, which
may be pre-set to lift its level to close to the standard (zero level or
line level) used within the console from this point on.
A second kind of input comes from outside the studio or from other
equipment within it, already at `line' level. Its entry to the console may
be through a jack-plug, and the signal will not necessarily be balanced.
So this signal is taken first through a small amplifier that balances and
feeds it forward to the channel. A mono channel is usually switchable,
so it can be used either for microphone or line input. An additional
small rotary `trim' control of about 15 dB serves either, adjusting its
level more precisely, so that the channel fader can be set to the zero
marked on its scale when it is making a full contribution to the mix.
Stereo line inputs (including the returns from artificial reverberation
devices) may also be fed to dedicated stereo channels, in which the
168
Monitoring and control
faders operate equally on both legs. This saves space on the console,
and treatments on the strip above the fader can be tailored more to
the demands of stereo. On some consoles, these fader units also have
a `remote-start' button that can be set up to replay recordings.
On both mono and stereo channels there should also be a polarity
control switch, often designated by the Greek letter phi, (for
`phase'). In mono, this simply switches the signal between the wires
that carry it, so that a stereo signal can be corrected if the wiring on
one leg is wrong (or it can be used to put the signal out of phase, for
some specially distorted effect). On a stereo channel, the polarity
control reverses one leg only, conventionally the left, or `A' leg.
[9.16] Flexible auxiliary send
switching
Used for public address, studio
feedback, AR `send' or any feed to
a remote location that needs a mix
that is different from the main studio
output. 1, Feed from channel
(or group) input. 2, Remote cut.
3, Channel (or group) fader. 4, Auxiliary
`send' switching. 5, Mono `send'
output fader. 6, `Send' output switch.
7, Stereo ganged output faders.
8, Stereo ganged output switches.
9, Pre-fade listen. 10, Channel out to
group bus or main control. 11, Output
to `mono send' buses. 12, Output to
`stereo send' buses.
Progressing up the strip, the signal passes through (or bypasses) an
equalizer and filters (described later) and a stage where other equipment may be plugged in. A `PFL' switch diverts a side chain to a
small speaker. Back on the main chain, there may then be a switch
which, if activated, permits the whole channel to be cut remotely:
this allows a `cough key' to be operated by a studio presenter, or a
`prompt-cut' when a performer is given instructions.
Still before the fader, there should be provision to split the signal and
send to auxiliary circuits ± which may include grouped feeds of
selected channels to AR devices, public address (to audience loudspeakers), foldback (to the studio) or feeds to anywhere else. These
may all require different combinations and mixes of the sources,
which must then be controlled separately, in another part of the
console.
After all this, the signal reappears back down at the bottom of the
strip, at the most important control of all, the channel fader. With all
the faders in a row along the near edge of the console, the balancer
can easily see and change their relative settings. In this position, they
can also be faded smoothly in and out.
Back up the strip, following the fader there will be controls
(described later) that place the signal within the stereo image. Finally,
a switch assigns the channel output to a group fader (usually stereo),
or via a subgroup fader (which may be mono or stereo).
[9.17] Routing matrix
A small panel at the top of each
channel strip on some mixers
designed for linking a separate
channel feed to a multichannel
recorder. The mass of press-buttons
looks more complex than it is: press
any button to send the channel output
to the selected track.
Consoles for use with multitrack recorders may have an additional
row of small modules, each with tiny buttons (for selection of send
or return) and also dedicated `dynamics' (compressor, limiter,
expander ± see next chapter) associated with each channel return.
At this level of complexity each channel is also likely to have some
kind of meter and this will be right at the top of the strip. Although
they are a long way up from their associated faders, this row of
meters, in line along the top of the console, gives an instant indication of where the action is and a good feel for its significance, as
well as all the relative levels.
The strip for each channel is often built up from a series of plug-in
modules. This strip layout does not necessarily reflect their true
sequence: it is arranged for operational convenience, as determined
by the console manufacturer, and in any case the actual wiring
makes repeated excursions to the patch bay. But it does make visual
Monitoring and control
169
sense, and when a source has been selected and plugged to a
channel, it is convenient to write its name or a microphone number
with a wax pencil or on a label in a line along the desk just above the
faders.
Group and master controls
For effective manual control, each of the mono and stereo channels
is assigned to a group, which is usually given a number. Colours can
also be used to identify groups. Letters, too, may be used, but at the
risk of confusion with `A' and `B' stereo legs.
[9.18] Groups
1±4, Bus-bars: these are actually
wires that cross all channels. In this
example two mono channels (5) are
both connected to bus 1, and the
stereo channel (6) is connected to
bus-bars 3 and 4, while nothing is fed
to bus 2. The output of each bus
(or stereo pair) is fed via group controls
to a main control. Auxiliary `sends'
also have their own buses and output
controls. 7, Group outputs.
The switches for group selection may be at the top of the channel
strips, or down near the faders; in some cases a display by the fader
shows which group it goes to. The selector may actually route the
audio feed to the group fader strip, or alternatively the group fader
may be part of a voltage-control amplifier (VCA), which acts as an
additional control on the individual channels. Whether or not the
audio signal passes through the group faders (or any fader on the
console), it is convenient to imagine that it does. Even though on
automated consoles many operations are remotely controlled, operationally the result is much the same.
Group faders require fewer elements on the strips above them ±
mainly stereo balance controls, routing and feed switches, and some
means of checking the signal at this stage: PFL and possibly, again,
a meter.
Analogue signals are added together at each stage by feeding them to
a common, low-impedance bus, from which they cannot return back
down through the other channels feeding the bus, because these
present a high impedance to the signal. The mixture is taken from
the bus by an amplifier that feeds it to the next stage.
[9.19] Main or master control
1±3, Stereo inputs from group faders.
4 and 5, Left and right main buses.
6, Ganged main control. 7, Limiter and
make-up amplifiers, also linked for
identical action on left and right
channels. 8, Zero-level line-up tone
input. 9, Feed to monitoring meters
and loudspeakers. 10, Stereo output.
The groups are in turn mixed (on a stereo pair of buses) to the master
fader, also known as the main gain control, then through a final
compressor/limiter to the studio output, where a monitoring side
chain feeds the main loudspeakers and meters. Meters, compressors
and limiters, and their role in volume control, are described in
Chapter 10.
The console will also usually have elaborate provision for treating
and routing the various auxiliary feeds or `independent main outputs' (taken from before or after the faders at each stage and passed
through their own submixers). Some of the uses for these are
described later. Alongside the main output there may be a second
independent or mix-minus (also called clean feed) output, taken off
before a final channel (perhaps for narrator or announcer) is added.
Plainly, the capacity of its console largely determines the limits of
what can be achieved in a studio. Individual balancers may become
proficient in operating and understanding the quirks of one console
to the almost total exclusion of all others, and conversion to another
170
Monitoring and control
may be accompanied by temporary loss of speed. This is an extra
reason why innovation is often disguised within relatively conventional layouts: `intuitive' layouts make for greater interchangeability. This is at a premium when working outside the studio using
equipment hired by the day. The balancer would usually prefer a
known console, even if it is a bit too big, but careful costing may
require that something smaller be selected. Brief catalogue descriptions will include the number of mono/line, stereo, group and
independent channels that are available, plus details of equalization,
auxiliary feeds, communications, and so on.
Remember, big is not necessarily beautiful: for some purposes, such
as modest film or video recordings on location, a small mixer with
just a few simple faders and no in-line facilities may be all that is
needed.
Faders
[9.20] Faders
Short- and long-throw conductive
plastic faders. These may also have
channel monitoring and group
selection controls on the fader strip.
Some faders used for line input may
also have remote-start buttons on
them.
A fader is essentially just a variable resistance or potentiometer,
sometimes called a `pot' for short. As a contact moves from one end
of its travel to the other, the resistance in the circuit increases or
decreases. At one end of the scale the resistance is zero; at the other
it is infinite (it moves right off the resistive surface or otherwise
interrupts the signal). The `law' of the fader is near-logarithmic over
much of its range, which means that a scale of decibels can be made
linear (or close to it) over a working range of perhaps 60 dB. If the
resistance were to increase according to the same law beyond this, it
would be twice as long before reaching a point where the signal is
negligible. But the range below 50 dB has little use, so here the
rate of fade increases rapidly to the final cut-off.
Faders on most professional equipment are flat and linear (as distinct
from older rotary or quadrant designs). The biggest `long-throw'
faders have an evenly spaced main working range (in decibels); where
less space is available, `short-throw' faders have the same distance
between 0 and 10 dB, but progressively less beyond that, and with
a more rapid fade-out below 40 dB. Music consoles designed to
work with two complete, parallel fader banks may have long-throw
faders on the nearest to the operator, with short-throw faders above.
On automated consoles, they have hidden motorized drives, so they
can move under the control of a computer. (On the most advanced
consoles other settings on the control strips are also automated.)
Each fader may be associated with several amplifiers. One of them
compensates for the action of the fader itself, so that the scale can be
marked with a zero (for 0 dB gain or loss) at a comfortable distance
from the top of the stroke ± in fact, with some 10 dB available above
the zero. The layout can be the same whether it directly controls an
audio signal that passes through it or, as a VCA fader, controls the
audio level from a distance. A minor refinement is that the knobs
may be in a range of colours, so that white might be used for channel
faders, blue for groups and red for the master.
Monitoring and control
171
The faders in most common use employ a strip of conductive
plastics. Carbon granules (in a complex layout, distributed to produce a `law' that is scaled in decibels) are printed on a plastic base
and heat-cured to render the surface smooth and hard-wearing. For
ease of operation and to reduce wear further, the slide makes only
light, multi-fingered contact with the resistive surface. High-quality
individual specimens may vary by up to a decibel from the average
(a range that is within the typical limits to human awareness). One
measure of performance is evenness of movement, the lack of any
`stick/slip' effect a user might detect when making light, fingertip
movements of the fader.
[9.21] Manual and motorized faders
Above: Manual operation only.
Below: This can be operated either
manually or by the wire from the
computer-controlled motor.
[9.22] Stops
This term originally referred to the
discrete jumps in level associated
with the spacing between studs on
older, stepped faders. These (with
many more studs than shown here)
were arranged at intervals of
approximately 2 dB, close to the limit
of human perception. The term
`stops' is still used, and the scale
markings on many of today's faders
correspond to them over their main
working range. Rotary faders are still
used for coarse pre-sets, or for
`trimming' over a narrower range, but
individually wired studs are now
rarely seen.
Simpler faders based on carbon-track or wire-wound resistors tend
to wear more rapidly and then make unpredictable crackles as the
slide contact is moved (as the owners of most domestic equipment
eventually discover). At one time, the only satisfactory alternative
was a fader with separate studs that the sliders engaged in turn, with
a fixed network of resistors between each successive pair of studs.
But it was difficult to make studs and sliders that always moved in
perfect contact, unaffected by dust, moisture and other atmospheric
contaminants, so these are now used only for coarse pre-sets.
In the days when they were used throughout a mixer, a rare crackle
or two was the price paid for being able to set levels precisely,
literally stop by stop in increments of 1.5 or 2 dB (for which the
term `stop' is still used). With the advent of stereo, these stops
provided exact matching between the two signals, though with some
danger of getting momentarily out of step when moving from one
pair of fader studs to the next: this could result in stereo image
wiggle or flicker.
Some of the simplest designs of fader have been used in low-level
mixers, i.e. those in which there is no preamplifier, or internal amplifiers compensating for the loss in the faders. But these are not suited to
mixing, where there is much use of faders, because as each additional
channel is opened it provides an alternative path for signals from the
others. In consequence, the faders are not independent in operation: as
fader 2 is brought up, there is a slight loss in signal 1, so that its fader
must be adjusted to compensate. Such problems are encountered only
in simple or non-professional mixers. A simple mixer for which it
might be suitable is that used for a subgroup of audience microphones
from which the combined output is fed to one stereo fader, so freeing
several channels with full facilities for other uses.
Stereo image controls
Rotary potentiometers are used at many points on the console where
space must be conserved. Several may be used (after the channel
faders) to construct or modify the stereo image:
.
Panpot, short for panoramic potentiometer. This steers or pans a
monophonic element to its proper position in the stereo image.
172
Monitoring and control
.
.
[9.23] Stereo controls
1, Pre-set control to balance output
between stereo microphone A and
B legs. 2, Ganged channel fader.
3, Image width control. 4, Channel
offset control (to displace the whole
spread image).
The input, fed through the moving contact, is divided in two and
fed to the A and B legs.
Image width control. This cross-links the A and B signals, feeding controlled and equal amounts of the A signal to the B
channel, and vice versa. As this cross-feed is increased, the image
width narrows. However, if at the same time the polarity is
reversed, the image width is increased.
Displacement control. This is a ganged pair of faders, one on
each path, and working in opposition to each other, so that turning the knob has the effect of moving the image sideways. This
control can also give apparent movement to a static but spread
source.
In a stereo channel the main pair of faders and the width control
might also be put into the circuit in a different way: by first converting A and B into M and S signals and operating on those instead.
When M and S signals are faded up or down there is no `wiggle';
instead, there might be minor variations in image width, but this is
less objectionable. In addition, image width can then be controlled
[9.24] Stereo signals
An electrical method for converting
A and B signals to M and S, and vice
versa. `M' is equivalent to the signal
that would be obtained by directing
a single suitable monophonic
microphone towards the centre of
the audio stage. It therefore gives a
mono output that can be used by
those without stereo loudspeakers,
although the balance between centre
and side subjects, and between
direct and reverberant sound, will
not necessarily be as good as for a
normal mono balance. `S' contains
some of the information from the
sides of the audio stage.
simply by changing the proportion of M to S. In particular, image
width can be increased without change of polarity. However, after
passing through its channel fader and width control, the M and S
signals must be converted back to A and B before passing through
the displacement faders.
Using microphones that produce the M and S signals directly can
eliminate one stage in this process. However, this implies a radical
change in working practices and the use of mixers that are at present
laid out primarily for AB operation. So it is usual to convert MS
microphone signals to AB and work with them in the usual way.
[9.25] Mono channels in stereo
A small rotary panpot in each channel
is used to place individual mono
components within the stereo image.
Where two monophonic channels with panpots are linked together to
form a stereo channel, image displacement and width are controlled
by the relative settings of the two panpots. If they both have the same
setting, the source has zero width, displaced to whatever position the
panpots give them. As the panpots are moved in opposite directions
Monitoring and control
173
from this point, the image increases in width. Full width is given by
steering one of the signals all the way left, to the A side, and the
other all the way right, to B in the mixed signal; this can, of course,
be done without a panpot at all.
A stage in the creation of surround sound employs two stereo panpots, one for side-to-side operation and the other for forward-to-rear.
By using the two together a monophonic source may be steered
anywhere within the four corners. For most purposes it is sufficient
to do this with the separate controls. But occasionally, for more
complex, continuously smooth movements, say of a character in
a play or an opera, or for special effects such as an aggressive insect
or a circling plane, signal pairs might also be fed through a joystick
control, or possibly controlled in two dimensions by a mouse or
a trackball.
[9.26] Panpot attenuation
A signal from a monophonic source is
split and fed through a pair of ganged
potentiometers: step by step, the
attenuation in the two channels is
matched (dotted lines). The apparent
volume will then be the same in all
available positions across the stage.
L, Left. 12L, Half left. C, Centre, etc.
Tracklaying, multitrack
recording and mix-down
In music recording studios, popular music is generally built up in
stages ± by tracklaying and overdubbing. An instrumental backing
might be recorded first on several tracks, followed by a soloist who
listens to a replay on headphones and adds the vocal, then a further
instrumental or vocal track might be provided by a musician who
played earlier. Treatment of various kinds may be tried, but not yet
confirmed.
The recordings are laid down using one of several multitrack recording systems. The longest established is the analogue multitrack
recorder: here the tracks are laid down side by side on broad tape
(24 tracks on 2-inch/52-mm tape or eight tracks on 1-inch/26-mm
tape) with separate stacks of heads to record and replay them. Next
came digital multitrack systems (24 or 48 tracks on half-inch/13-mm
tape). A cheaper competitor is the multitrack digital module
(MDM): each MDM has eight tracks on an inexpensive domesticstyle videotape, and the modules are stacked to provide anything up
to 128 recording tracks. Eight is a good number for 5.1 surround
sound mixdown: 5 plus 1, with 2 for a separate stereo mix.
[9.27] Joystick panpot
On some consoles, in addition to the
rotary panpots in each channel, any
selected signal can be fed via a small
joystick control. The display above
shows its panned position in 5.1.
Any of these recording systems can be traded in for (or backed up
by) hard disk on a PC or Mac computer, which costs even less, and
also removes the obligation to think in multiples of eight. Alternatively, a removable hard drive is housed in a small rectangular box
with SCSI connector pins at the back. A recording made in one place
can be taken away and plugged directly to the input of an audio
editing system (which can be anywhere) and later returned to
another studio. This whole pluggable SCSI box is usually called
a `scuzzy' for short.
On many consoles each channel has an array of numbered selector
buttons, each of which sends a feed from it to a particular track
(or stereo to pairs of tracks). A trial mix is obtained during the session
174
Monitoring and control
(so producers and others can get a feel for what is being achieved)
and this may also be recorded for reference: the levels and settings
set for all channels are noted ± or better, stored in the desk by
computer.
A console that can store information generally has `write', `read'
and `update' modes, and may also have additional `write' and
`update' buttons associated with each fader (moving fader systems
normally have touch-sensitive fader knobs) and in more sophisticated consoles, the EQ and panpots, etc., are automated too. The
initial balance is made with the mixer set to the `write' mode. When
the tape is replayed with the mixer in the `read' condition, all the
earlier fader settings and changes are reproduced (irrespective of
their current positions). If while the mix is being read a `hold' button
is pressed, replay continues but with the controls unchanged from
that point on: this could be used (for example) to get rid of a fade-out
on the end of the initial mix.
Except in the case where the computer also moves the controls, the
mix is updated as follows: first, all faders are set to zero or to some
nominal setting which becomes the `null' level. Then, after the
relevant channels have been set to `update', the tape is replayed
and any changes from the null position (and their variation in time)
are fed to the data store, changing the selected parts of the mix by the
amounts indicated. Alternatively, the whole of the previous control
sequence for a channel can be rejected and written afresh by using
the `write' button instead of `update' and operating the fader normally. Also associated with each fader is a grouping switch. If
several faders are dialled to the same number to form a subgroup,
the `write' and `update' instructions for any one of them operate on
the control data for all members of the subgroup.
This mix-down generally takes place, after the departure of the
musicians, in a post-balancing session, perhaps even on another
day. At this stage the recording systems of choice (again keeping a
backup) might include DAT, magneto-optical (MO) or hard disk. If
it is destined for CD, it will go through a digital audio workstation
(DAW) anyway.
In radio and television, where time is more limited, multitracking is
more often used just to allow for additions, minor improvements or
second thoughts.
Control console checks
Operating a console is like driving a car: easy, given training and
experience ± and provided that the car starts and is in working order.
With hundreds of faders, switches and routing controls to choose
from, if you cannot hear what you think you should be hearing it is
as well to have a systematic strategy for recovery, which should
include the apparently obvious. Consoles vary, but a basic troubleshooting checklist might start like this:
Monitoring and control
.
.
.
.
.
.
.
175
Power. Is the console switched on? Is power reaching it? Has the
supply failed?
Microphone or line source. Is it working? Is it fully plugged in?
If in doubt, try another (but not one that takes power from the
console).
Jackfield. Is the routing correct? Is the path intercepted (and so
cut)? Are the jacks fully in? Has moisture condensed into it?
(This is among the hazards for equipment taken away from the
studio.)
Channel input selector. Is it correctly set to microphone or line?
Routing within console. Is it going to an audio group or VCA or
as an independent direct to the main fader?
Faders. Are all on the chosen routing open? Open them all to
`0 dB'.
Solo, cut and AFL switches. Has the signal been interrupted by a
switch on another channel?
Make up your own checklist, or use one that others have written for
that studio. This might also contain notes on performance limitations
and how to make the best of them. Keep a fault, maintenance and
repair log, and in a studio used by several operators a suggestion book.
With the growing complexity of studio equipment, maintenance is
increasingly a matter of module replacement, of dealing with `black
boxes' rather than understanding the detail of their internal circuits,
and how to investigate and repair them on the spot. But it is essential
to be aware of operational problems, such as any switch that makes a
click on the line, so that it can be avoided when on-air or recording.
The user should know how and be prepared to route round any
module that fails, especially on a live broadcast. It helps if operators
understand signal pathways through the console, are familiar with
common deficiencies in quality, and know how to localize and
describe them.
Monitoring sound quality: noise
Before we can begin to use the console and its monitoring system
creatively, we must first know how to judge sound quality ± if only
to distinguish between artistic and technical faults. When there is a
fault, there is liable to be a frustrating and potentially expensive
delay until it is tracked down to its source and cured or isolated.
Some of the `traditional' problems associated with analogue audio
are described below. Digital audio bypasses some of these, but
introduces others of its own: some digital problems are described
in Chapter 16.
Sound degradation problem number one is noise, which for our
present purpose may be regarded as random sound of indefinite
pitch (there are many other forms of `unwanted sound', but they
can each be treated separately). Irregularities in the structure of
matter cause noise: the particles that are magnetized (or missing)
on tape or disk, roughness in the wall of an old record, the granular
176
Monitoring and control
structure of a resistor, the random dance of individual electrons in
semiconductor or metal; all produce some noise in a recording or
broadcast. They are heard as crackling, buzzing or hissing, and must
all be kept at low level compared with the audio signal. We need a
good signal-to-noise ratio.
Noise becomes serious when components before the first stage of
amplification begin to deteriorate; at a later stage, when there are
more drastic faults (e.g. a `dirty' fader or a `dry' joint), where
vibration reaches a transducer, where there are long lines, or where
radio transmissions are received in unfavourable conditions. To
combat noise, in recording or transmission systems where noise
levels increase with frequency in a predictable manner, the highfrequency signal is increased (pre-emphasized) before the process
and reduced (de-emphasized) after it. For radio or television, these
systems require national or international standards, so that professional and domestic equipment produce complementary effects. One
advance in noise reduction came with the switch from amplitude to
frequency modulation of broadcast audio signals, another when
signals could be passed along the audio chain in digital rather than
analogue form. Even so, a great amount of analogue equipment
remains in use.
For the recording industry a range of commercial noise reduction
systems is widely used, with different standards applied to domestic
recordings, professional audio, or optical or magnetic recordings on
film. Most divide the sound into a number of frequency ranges and
apply treatments to each, together with a control signal that allows
the reproducer to decode it.
In one noise reduction system that is used in professional studios for
analogue recordings, the signal is split by filters into four frequency
bands: 80 Hz low-pass; 80 Hz to 3 kHz; 3 kHz high-pass; and 9 kHz
high-pass. Each band is separately compressed, so that very-lowlevel signals (40 dB down) are raised to 10 dB, while signals from 0
to 20 dB down are virtually unaffected. The recovery time (the time
during which the level returns substantially to its original state) is
1±100 ms, depending on the signal dynamics. There are also crude
volume limiters in each channel: these are simple peak choppers.
A matching network is used to reconstitute the original signal
on playback.
Such systems rely on the idea that noise that is similar in frequency
to the louder components of the signal is masked by it, while noise
that is widely separated in frequency from the louder parts can be
reduced in level: the improvement is therefore partly real and partly
apparent. As the system produces a coded signal that must then be
decoded, it cannot be used to improve an already noisy signal. It is
widely used to improve the quality of tape recordings, where an
increase in the signal-to-noise ratio is used partly to reduce hiss, hum
and rumble, and also by lowering recording levels a little, to produce
less distortion and to lower the danger of print-through. It is particularly valuable for multitrack recordings, but can in principle be
used around any noisy component or link. The signal is, of course,
unusable in its coded form: characteristically it sounds bright,
Monitoring and control
177
breathy and larger than life. (The noise reduction systems used on
domestic equipment are much simpler. For further information on
these and other analogue noise reduction systems, refer to specialist
literature.)
Digital recording offers a 90 dB signal-to-noise ratio without further
treatment.
Hum
[9.28] Earth loop (ground loop)
In this example: 1, Microphone
channel. 2 and 3, To earth by two
separate paths. 4, The loop is
completed invisibly, through the
common earth. 5, An unrelated mains
cable, crossing through the loop,
creates an electromotive force that
generates a matching alternating
current in the loop. 6, In this worst
case, a preamplifier selectively
boosts only the forward half of each
cycle. 7, This amplifies the hum and
generates rich harmonics, of which
the first overtone (second harmonic)
is particularly strong and audible. If it
is difficult to find and eliminate the
additional path to earth, it may be
necessary to filter the signal, giving
particular attention to the second
harmonic (at 100 or 120 Hz).
A noise that does have definite pitch is hum ± often due to the mains
frequency and its harmonics getting into the audio signal. Mains
hum may be caused by inadequate smoothing of an electrical power
supply when rectified from alternating to direct current, or by
inadequate screening on wires carrying audio signals, particularly
those at low level, or in high-impedance circuits, such as output from
condenser microphones before the first amplifier. Magnetic microphones can be directly affected by nearby electrical equipment, such
as mains transformers or electric motors. Hum may also be caused
by making connections by separate paths to a common earth, which
creates an earth loop that picks up signals from power lines that pass
through it.
The second harmonic of mains hum is generally the most audible, so
most hum filters just get rid of that (at 100 or 120 Hz). More
elaborate `notch' filters take out narrow slices of a wider range of
harmonics without substantially affecting music. Some `plug-in'
digital signal processors have more sophisticated programs that
include this; they also tackle ventilation noise, guitar amp buzz, hiss
from archive recordings and have extra programs for clicks and
pops. (But note that if applied too aggressively, they introduce
artifacts of their own.)
Hum can be a problem when recording popular music from public
performances, where feeds are split from the system set up for the
stage foldback loudspeakers. To identify the source of hum quickly,
fit the earth (ground) leads for each circuit with a switch so that each
can be individually checked. Where sound is combined with vision,
lighting (or their dimmers) is the usual suspect. A full rehearsal will
show up danger points ± or ask for a quick run through the lighting
plot to hear what it does to open microphone circuits. Portable humdetection devices are also useful when setting up any temporary rig
that involves a lot of audio and lighting circuits.
Distortion
Distortion should not be encountered in modern systems that are
working well, but was a serious problem in the days when successive
components were less well protected. If at any point in a broadcast
178
Monitoring and control
or recording chain the volume of an analogue signal is set too high
for any stage that follows, the waveform is changed ± the peaks are
flattened, creating new harmonics. However, most are added at
frequencies that are already present in much sound, and most music.
The introduction of this harmonic distortion is therefore partly
masked by existing sound, and even where it is not, it does at least
have a pseudo-musical relationship to the wanted sound. Speech
sibilants, which do not have a harmonic structure, bring it out
strongly. Note, however, that human hearing is itself subject to
distortion at high acoustic intensities.
Small amounts of distortion may be more acceptable than a high
noise level: 2% harmonic distortion is generally noticeable and 1%
represents a reasonable limit for high quality, but with up to 3% on
some peaks.
Unduly high distortion may be due to faults in equipment, such as
where the moving part of a transducer has worked loose (e.g. a
ribbon or moving coil hanging out of its magnetic field), or errors
in their use, such as if the wrong bias is used when recording on
magnetic tape.
[9.29] Harmonic distortion
The dotted portion of the signals is
chopped off, flattened.
A, `Bottoming' causes part of each
cycle to be clipped. B, Both peaks are
flattened, as excessive swings in
each direction cannot be followed by
the signal in some item of equipment.
In both cases the distortion produces
many overtones, at exact multiples of
the frequency of clipping.
Intermodulation distortion is far worse than harmonic distortion: it
occurs when two frequencies interfere with each other to produce
sum or difference products. It should also be rare: good sound
equipment should be designed to avoid intermodulation effects like
a plague.
Distortion may arise at the start of the chain, at points that are,
in theory, under the control of the balancer. Some air pressures
(perhaps caused by wind, or by plosive consonants in close speech)
may move a microphone diaphragm outside its normal working range.
Or microphone output may be too high for the first preamplifier.
Amplifiers have a linear response to changes in volume (i.e. output
is proportional to input) only over a limited range. The sensitivity of
professional microphones is related to the volume of expected sound
sources in such a way that they mostly feed signals to the mixer at
about the same range of levels. Exceptions are bass drum, brass
and also electric guitar loudspeakers that are set very high. If
following preamplifiers are of fixed gain, a `pad' (a fixed-loss
attenuator of, say, 12 dB) may be required between the microphone
and amplifier. Alternatively, use a microphone of lower sensitivity
or one with a head amplifier that can be switched to accept different
input levels.
Another example of this form of distortion may occur in very close
piano balance. In some popular music the piano becomes a percussion instrument, with powerful attack transients. The brief, high
volume of these transients may not be obvious from any meter that
takes time to respond, but this too is liable to distort.
Note that if a digital signal is clipped, this (unlike analogue clipping)
immediately results in unacceptable loss of data. To avoid this,
digital recordings therefore generally require either 10 dB of clear,
extra `headroom' or consummate care if higher levels are permitted.
Monitoring and control
179
Crosstalk
On stereo, crosstalk is a defect to listen out for. It is measured as the
level at which a tone at a particular frequency in one stereo leg can
be heard on the other. But what is regarded as `unacceptable' seems
to vary. For example, for broadcast tape recordings the limit might
be set at 38 dB at 1 kHz, rising to 30 dB at 50 Hz and 10 kHz.
These figures compare favourably with the 25 dB (at middle frequencies) available in the groove of a stereo vinyl disc, but are
extremely poor compared with digital recordings. In a good mixer,
the specification will be much higher, perhaps requiring a separation
of more than 60 dB ± and one console offers better than 75 dB at
15 kHz.
Subjective quality checks
Checking the technical quality of work in progress should be automatic and continuous, so that any fault (and with experience an
immediate diagnosis of its probable cause) simply seems to jump
out at you. But there are other circumstances, in principle just as
important, in which faults may be there for all to hear, but none of
those responsible seem to notice. This may happen when the product
is no longer just a single work, but a whole body of output, such as a
broadcast channel.
[9.30] Subjective scales for quality
assessment
Scale 1
1.1
1.2
1.3
1.4
1.5
1.6
Sound degradation
Imperceptible
Just perceptible
Audible but not disturbing
Somewhat objectionable
Definitely objectionable
Unusable
Scale 2
2.1
2.2
2.3
2.4
2.5
2.6
Overall quality
Excellent
Good
Fair
Fairly poor
Poor
Very poor
Judgements of sound quality are personal and subjective, but in such
cases it may still be helpful to try to apply standards that are common
to any group of people who work together. A broadcaster may
attempt to do this by setting up a system of observations that might
cover general sound quality (including background hum, noise or
hiss), change of sound level at programme junctions, and interference. One of the items in this last category is the talkback induction
sometimes heard on radio, but even more often on television because
of the continuous stream of instructions and information fed from the
gallery to the studio floor. And ± this may seem a lost cause, but add
it in anyway ± most of the public would like a less intrusive match
between the sound levels of programmes and commercials.
In one example of such an assessment system, sampling was done
for periods of 5 minutes on a wide range of material, including
productions as they were transmitted (both before the transmitter
and off-air), studio output (but taking into account conditions that
may exist in rehearsal before all adjustments have been made),
material incoming from other areas (by landline, radio link or
satellite) and all recordings.
Because they cover such a wide range of possible defects and because
others might appear unexpectedly, the most useful scales are likely to
seem highly subjective. The divisions in the scales to the left may
seem coarse, but they are readily assessed and easily understood.
Chapter 10
Volume and dynamics
There are three kinds of volume control. The first uses meters to
control overall level. For the second, to control internal dynamics, we
need loudspeakers and our ears. The third employs compressors and
limiters and is fully automatic but may still require monitoring by ear.
The first application of volume control is purely technical. Meters
are used to keep the signal within limits that allow it to pass with
minimal unwanted changes through equipment, along lines and
links, to reach broadcast transmitters or recording media. If an
analogue level is too high, preamplifiers will overload and distort,
tape recording media will be saturated, and (even worse) any attempt
to convert analogue to digital will cause signal clipping, in which the
data is severely damaged. If the signal is too low, it has to compete
with noise, and even normally robust digital signals can suffer. For
checking maximum volume, loudspeakers are no help at all, as their
settings are arbitrary, but they are essential for listening to changes
in volume (dynamics).
Ensuring that what the audience should hear lies between acceptable
maximum and minimum volumes may also require reduction of
dynamic range, by either manual or automatic compression. Fortunately, many people listening at home prefer a reduced range. For
this, it is the meter that gives little help: aesthetically satisfying
comparative volumes can be assessed only by ear. Meters are a
guide, but ears must be the judge.
Automatic compression can be used to control some kinds of sound
reasonably well ± but not all. Automatic control lacks intelligent
anticipation: it would, for example, ruin Ravel's Bolero. In most
music, subtle dynamic control by the performers is important to the
artistic effect and no compressor will do an adequate job unless it
can be set for the specific item. Another kind of problem with
relative volumes is that different control characteristics are required
where several types of output are heard in succession.
Volume and dynamics
181
Specialized uses for compressors and limiters include short-wave
radio, where a sustained high-signal level is necessary to overcome
static and other noise; also certain other types of high-intelligibility
radio, where a limited range of output is broadcast to a maximum service
area for a given transmitter strength. In addition, they serve particular
purposes within programmes, e.g. as an element of pop music balance.
Meters
Although a meter cannot be used to judge relative levels, such as that
of one voice against another or of speech against music, it is nevertheless an essential item of studio equipment. Its uses are:
.
.
.
.
.
to check that there is no loss or gain in level between studio and
recorder or transmitter;
to avoid levels that would result in under- or overmodulation at
the recorder or transmitter;
to compare relative levels between one performance and another;
to check that levels meet prescribed values for good listening;
to provide rough visual checks on the relative levels (and correct
polarity) of stereo channels and mono compatibility.
There are two main kinds of meter for analogue audio, operating on
completely different principles. One responds to the audio signal
logarithmically (as our hearing does); the other does not.
A peak programme meter (PPM, or `peak meter'), well established
in European practice, is basically a voltmeter that reads logarithmically over its main working range. In order to give prominence to the
all-important `peaks' that may be of short duration but vulnerable to
overload and therefore distortion, it has a rapid rise characteristic
(`time constant') and slow die-away that makes it easy to read. The
PPM originally developed for BBC use has a rise time constant of
2.5 ms, giving 80% of full deflection in 4 ms (our ears cannot detect
distortion of less than this duration), and a die-away time constant of
1 s, a fall of 8.7 dB/s. Other European meters may have different rise
and fall times and face-scales from those favoured in Britain.
[10.1] Peak programme meter
(PPM)
The BBC standard PPM is widely
used for analogue metering in
Europe. Markings are white on black.
Each division is 4 dB; `4' on the meter
is a standard `zero level' and `6' is full
modulation, above which distortion is
to be expected. The needles may be
set up to show either A and B or
M and S signals.
Placed prominently on the middle of the console, the original mono
PPM had a single needle that, for visual clarity, moved over a black
face with white markings. For stereo this was replaced by two
meters, each with two concentric needles. One meter is for the A
and B signals, in colours that follow the nautical `red-port-left' and
`green-starboard-right' convention. The second shows M and S
signals, conventionally with white and yellow needles (and not
always easy to distinguish). Because M ˆ A ‡ B it normally reads
higher than either: for a source in the middle it is 3 dB higher.
Because S ˆ A B, it is normally less than M. Unless there is phase
cancellation, S is to the left of M, so if S swings to the right it
suggests that a leg is wired with its polarity reversed: this is an
immediate visual check for the fault. (But beware: in some consoles
the functions of the two PPMs can be swapped over.)
182
Volume and dynamics
PPMs are used, as the name suggests, to protect against analogue
peaks that may cause distortion. However, although the rise is rapid,
peaks of very short duration will not be fully registered: the true
peak may be up to 4 dB higher. Also, if digital recordings exceed the
`100%' level, the damage due to loss of data is so severe that this
`maximum coding level' is taken as an absolute limit. To guard
against excessive unpredictable peak levels and allow for any
under-reading, an alignment with peaks 10 dB below that ± 10 dB
`headroom' ± is adopted for digital recordings. On a DAT recorder,
for example, peaks should be this much below the `0' that marks
maximum coding level. The loss in signal-to-noise ratio is a fair
trade against the danger of digital clipping.
An exception is made for CDs, which are used more for the distribution
of digital recordings than as the original recording medium. CD
mastering is controlled very precisely. Digital editing systems allow
very accurate measurement of the highest peaks, so when the recording is `burned' to CD-R this can be used to ensure that even
momentary peaks do not exceed the maximum coding level. By
using this `normalization' process, a CD-R disc can have a level
similar to that of a commercial CD. If the level is read on a PPM its
apparent headroom will be 4 dB (to allow for the meter's slower
response).
The alternative (again designed for analogue sound) is the VU meter.
Widely used in American equipment, this is marked with two scales:
one shows `percentage modulation' (used more in broadcasting) and
the other reads in decibels. The VU meter is fed through a dry
rectifier and ballast resistance, and (unlike the more complex
PPM) may draw from the programme circuit all the power needed
to operate it. For cheaper models the ballistic operation may allow a
needle to peak substantially higher on music or speech than on
steady tone. In contrast, a more expensive, studio-quality VU meter
overshoots only by a little when a pulse signal is applied (in which it
compares favourably with a PPM). But a time constant of 300 ms
prevents transient signals from being registered at all, and the meter
under-reads on both percussive sounds and speech, so at its `100%
modulation' level these may be distorted.
[10.2] VU meter
Widely used for analogue sound
metering in America. The lower scale
indicates `percentage modulation'
and the upper scale is decibels
relative to full scale.
Though it may have a linear calibration in `percentage modulation'
(the rectifier characteristic should ensure this), more than half of that
scale is taken up with a range of 3 dB on either side of the nominal
`100%'. Little programme material remains consistently within such
a narrow range, so the needle is likely to register either only small
deflections or flickers bewilderingly over the full range of the scale.
For maximum volumes, continuous sounds (including most music)
may read `100% modulation' on the VU meter; however, for staccato sounds such as speech or drum taps a true maximum is given by
an indicated 50%, because the meter does not have time to register a
full swing to the 100% mark on the louder individual components of
speech.
Some balancers who have always used a VU meter say they like the
way its reading seems to relate more to average loudness than a PPM
does. For a signal that is supposed to stay close to full modulation
Volume and dynamics
183
they try to keep the needle close to the nominal zero in a range
somewhere above 5 and not above ‡3. For some kinds of music
the roughening of tone caused by a slight overrun can even be
regarded as beneficial: the term `gutsy' has been used for this effect.
A drum-kit that may already be overmodulating at `VU 0' sounds
even more dramatic (different, certainly) when the needle is `pinned'
over to the right. VU meters seem better suited to the record
industry, which has more time for experimentation, than they are
to radio and television, where excessive peaks are more of a menace.
Note that only those VU meters that conform to the relevant American
standard can be used to compare programme material. Many meters
do not meet this standard. Some bargraph meters can be switched
between peak and VU meter calibration, or (for the best of all
possible worlds) have both, side by side.
[10.3] Simple LED bargraph
As a rough guide to level (or overload)
in an individual channel or group,
a simple scale with LED segments
may be adequate. Here each division
below zero (maximum volume) is
3 dB.
A meter designed (or set) primarily for digital signals has the zero
(to mark maximum coding level, above which `clipping' will occur)
near the right (or top) of the scale. The main working range is well to
the left of (or below) that, 10 dB below if recommended `headroom'
is allowed for safety. Unlike the PPM, this meter shows instantaneous peaks, but still with a decay that is slow enough for the eye to
register easily.
On a radio control console, the meter (of whatever type) is likely to
be either centrally placed or roughly in line with the main control
fader; a television sound output meter is mounted in line with the
transmission monitor and, to avoid eye fatigue, in much the same
visual plane.
Line-up tone
Meters are used to establish common levels at all stages in the chain
by using a pure tone, usually at 1000 Hz and of known volume, as a
reference. Organizations (such as the BBC) with a central control
centre for signal routing will have a reference tone source at that
point, and most consoles and some other items of equipment such as
recording machines also have their own. A studio will therefore
often have the choice of using the central source or its own local
tone ± which should, of course, be the same. This always registers
`4' on a monophonic PPM (or as the M reading in stereo, with
nothing at all on S). Equal stereo A and B signals should each read
31¤4, i.e. 3 dB below `4'. Tones from other sources will be faded up
until they reach this same reading, to establish the incoming zero
level; any losses on the incoming line should be made good at the
input to the console.
The standard used is 1 mW into 600 , and for an analogue signal is
equivalent to 40% modulation at the transmitter (so that 100% is
about 8 dB above this). At all stages in the chain the tone should give
the same steady reading on all meters, and when a recording is made
(other than for public distribution), reference tone is recorded on it at
184
Volume and dynamics
the appropriate level for the equipment and recording medium, so
that the process can be repeated when it is replayed.
Line-up tone is also used to set a common level on recordings that
are to be used in broadcasting. Different standards are set for
analogue recordings, digital recordings such as DAT that require
more headroom in order keep clear of the digital maximum coding
level, and the anomalous CD-R, for which the (known) recorded
level may be higher, in order to match that on a commercial CD.
Documentation kept with recordings should be clear on what alignment has been used.
As a practical guide to consistent standards, the following are
maximum programme and line-up levels ± in general, following
AES/EBU recommendations ± as adopted by the BBC and its
suppliers for the most common media. As usual, the meter is a
PPM on which a reading of `6' represents maximum programme
level (`100% modulation') and `4' (the standard A ‡ B ˆ M zero
level tone) is 8 dB below that.
Open-reel analogue tape recordings should have 20 s of 1 kHz
line-up tone at the start. For full-track mono this will be at zero
level. For twin-track stereo this corresponds to a level 3 dB lower on
each leg, but M (the sum of A ‡ B) will still read `4'. Follow this
with 10 s of recorded silence, then 4 s of yellow leader immediately
before the programme. At the end put 3 s of red or red-and-white
trailer. If any additional material (such as retakes) is to go on the
end, leave 2 minutes of recorded silence, then mark the extras with
yellow spacers only ± not red. Note that even though a standard
alignment has been used for line-up, its level should still be noted in
the documentation (and on the box).
Analogue multitrack tape should have, on all tracks in parallel,
1 minute of 1 kHz tone at 4 dB below peak level, then 1 minute each
of 10 kHz and 100 Hz, both at 14 dB below peak level. If a noise
reduction system is used, this requires more tone (check the instructions for the system used). Follow this with 30 s of recorded silence,
then the programme.
On DAT, for which the maximum A ‡ B programme level is taken to
be 10 dB below the full scale, stereo (A ‡ B) line-up tone will be set
at 18 dB below full scale (or the individual A and B legs will be
21 dB below full scale). On replay these will be matched to `4' (or
`31¤4'). If the sampling frequency is set at 44.1 kHz (16-bit) to match
that for CD, it should also be possible to disengage `copy prohibit' in
order to allow for an unlimited number of copies to be made. To
allow quick-start replay, time code should be recorded continuously
from the start of the tape.
Unless otherwise required, it is good working practice for a DAT
tape to begin with 30 s of recorded silence, 30 s 1 kHz tone (set at the
above level), another 30 s recorded silence, then a start ID: this will
be the first and ideally only ID on the tape. The programme should
follow half a second after the ID: this allows the ID to be used to set
it up (unambiguously) for replay, and can also be used for quick-start
Volume and dynamics
185
or memory-start operation. Note that the programme (especially
music) may begin with a second or two of atmosphere.
If further items must go on the same DAT tape, allow 15±30 s gaps,
followed by consecutively numbered start IDs, but without any more
line-up tone. If there is an `auto-ID' function, switch it off, to avoid
any confusing double IDs. Finally, set the cassette `record protect' tab.
Digital multitrack tape also follows the rule of allowing 10 dB
headroom on the actual recordings ± but start the tape with 1 minute
of 1 kHz tone at zero level (i.e. 8 dB below the standard peak level),
then 30 s of recorded silence before the programme.
On CD, for which the maximum A ‡ B level on a PPM needs to be
4 dB below full scale, stereo (A ‡ B) line-up is set at 12 below full
scale (or the A and B legs at 15 below full scale). On replay these
match up to `4' (or `31¤4'). CD-R is recorded with 30 s of 1 kHz lineup tone after 2 minutes of recorded silence at the end of the disc
(so an 80-minute disc can hold only 77 minutes of usable material).
Documentation should make it clear whether normalization has
been used to lift the level to the maximum available.
With MiniDisc (MD), before recording, select for analogue or digital
input and follow instructions in the manual for other pre-sets. There
is a mode on the machine to name the item (but if it is fed by
computer, it is quicker to use its keyboard). Check the level of the
material to be recorded: this should have 10 dB clear headroom
above peaks, as for DAT. No additional line-up is necessary. Set
the `record protect' tab.
If copying to MD from a digital edit (for example, in one common
use for MD, of a broadcast trailer) run the play-out, then about a
second before the required material crosses the play cursor, switch
MD to record. For each successive track, press `record' again to
insert a new track number. At the end, switch back to `standby' and
wait a few seconds for the machine to complete writing additional
information on the disc. There are procedures for trimming recordings (`topping' and `tailing') if required. Machines vary, so check
the instructions.
Note that, in addition to line-up, broadcasters generally ask for
(or by contract, require) recordings to conform to rigorous technical
standards with specified documentation, and will usually supply
information on this on request. It is plainly in everyone's interest
to ensure the best available quality where this is possible. For some
suggestions (or requirements) about documentation, see Chapter 20.
Multitrack meters
In music recording studios, multitrack techniques ideally require
separate metering of each channel that is being recorded. For this,
there will be in-line displays on all channels, and meters may also be
supplied for auxiliary `sends'. Mechanical PPMs are unnecessarily
186
Volume and dynamics
complex, expensive and also difficult to follow in such numbers. If
needle-pointer displays are used, the simpler VU meters are adequate,
indicating which channels are not only faded up but also passing a
signal. In America, these have sometimes been arranged in pairs, one
above the other for each stereo channel, all across a large mixer.
A clearer option is bargraphs. These are vertical strips ranged side by
side, presenting a graphic display that is much easier to follow. With
processor control, they can be switched to show a variety of VU or
PPM recordings, with stereo in AB or MS pairs, and may also have
overload or `hold' options. In one example, they can be switched over
to show a frequency spectrum, analysed as one-third octave bands.
A bargraph display may take one of several forms. A strip of either
LED or liquid crystal segments is perhaps the simplest. More flexible, in that its brightness can be controlled, is the plasma bargraph:
this becomes brighter in the `overload' range. But it also takes more
power and costs more. A third type, displayed on a small screen, just
simulates the bar.
For some channels that do not justify the cost of any kind of meter, an
`overload' light may be all that is needed ± to give the simplest
possible warning that the upper limit has been breached, so `back off!'
Controlling stereo
Where the stereo broadcast is on FM, control is easy. Since the
deviation (size) of the signal does not exceed the greater of the
A and B channels, all that is necessary to avoid overmodulation is
that these do not exceed their normal limits. The same is true for
analogue recordings where the A and B channels are kept separate.
For systems like these, it is necessary only to keep the separate
A and B signals below the standard `full modulation' mark.
In broadcasting, however, several problems may arise:
.
.
.
[10.4] Bargraphs
1, Mono. 2, Stereo (or any pair of
channels). The simplest has LED
segments. The more complex, bright
plasma display can be programmed
to offer a variety of calibrations,
including the standard PPM and VU
response. Above maximum volume
(here ‡8 on the PPM), the plasma
glow is brighter still, to signal
overload.
in a large network some of the transmitters may be in mono;
some of the listeners to the stereo broadcast have mono receivers;
some items on the stereo service may be broadcast in mono only.
For practical purposes it is assumed that the average level for A ‡ B
is 3 dB higher than A or B on its own; accordingly, the stereo and
mono parts of the system are lined up with a 3 dB difference in level.
However, what happens to a dynamic signal is a little more complicated. There would be no difficulty if the signals added simply, so
the A ‡ B maximum volume really were never more than 3 dB
above the separate maximum volumes for A ‡ B, but in fact this
happens only for identical signals, such as tone or a source in the
centre of the sound stage. If A and B are both at the same volume but
are actually different sounds, their sum varies from ‡0 to ‡6 dB. So
where A and B are at maximum volume the M signal may be
overmodulating by 3 dB. This should therefore be controlled for
the benefit of mono systems.
Volume and dynamics
187
On the other hand, if the signal is being controlled primarily on the
M signal, the full stereo signal can overmodulate. This happens
when A ‡ B is at its maximum and the whole of the signal is in
the A or B channel, which is then 3 dB over the top.
A visual comparison of the M and S signal levels is less crucial, but
can give useful information about the width of the stereo image. If S is
close to M, the stereo is wide and a significant amount of information
may be lost to any mono listener. If S is 3 dB lower, the stereo content
is modest, and the width of the image is relatively narrow.
Programme volume: relative
levels
[10.5] Switchable bargraph
Here, zero level on the PPM scale is
indicated by 0 not 4, as on the older,
electromechanical PPM. Maximum
volume is therefore ‡8, which
corresponds to zero on the VU.
[10.6] Suggested PPM peak levels
(dB relative to 100% modulation)
Talk, discussion
News and weather
Drama: narration
Drama: action
Light music
Classical music
Harpsichords and
bagpipes
Clavichords and
virginals
Announcer between
musical items
0
6
8
0 to 16
0 to 16
0 to 22
or lower
8
16
4 to
8*
*Depending on type of music
These recommendations predate the
commercial era in which promotional
material is habitually given maximum
volume, but they do still offer guidance
on what listeners might really prefer.
A peak programme meter can be used to suggest or check levels. For
example, if ordinary talk or discussion is allowed a normal dynamic
range with occasional peaks up to the maximum, then newsreaders
(who generally speak very clearly and evenly) usually sound loud
enough with peaks averaging about 6 dB less.
A whole system of such reference levels can be built up; the table
shows specimen PPM peak levels for speech and music that experience
indicates are about right for listeners hearing radio in reasonably
good conditions. Some more unusual instruments are listed as
examples of how far some sound sources can depart from `average'.
Even if such a list is adopted as a general guide, it does not solve all
the problems posed by junctions between speech and music or, on
the larger scale, between one programme and another. For radio
stations with a restricted range of programming, there is a limited
number of types of junction, and it is fairly easy to link them without
apparent jumps in level. But for a service that includes news,
comedy, magazines and discussion programmes, religious services
and various kinds of music, problems of matching between successive items can be acute ± and there is no rule-of-thumb answer. It is a
question of judgement, of trying to imagine what the audience will
hear when listening to your programme for the first and (probably)
only time. Remember that your own perceptions are coloured by
long acquaintance with the material: familiarity breeds acceptance.
Remember, too, that the audience may not agree that your masterpiece should be displayed more prominently than anything around it.
A further slight complication is suggested by the results of a study of
listeners' preferences for relative levels of speech and music. Taking
a large number of pieces in different styles (but excluding modern
pop music) and with different linking voices, the results were:
.
.
speech following music should be (on average) 4 dB down;
music following speech should be (on average) 2 dB up.
This apparent contradiction is easy enough to resolve: it just means that
announcements linking music items should be edged in a little. The
aesthetic use of faders for links like these is discussed in Chapter 14.
188
Volume and dynamics
Maximum volumes
[10.7] Preferred maximum sound levels
(dB, referred to 2 10 2 Pa)
Public
M F
Symph. mus. 78
Light mus.
75
Speech
71
78
74
71
*M, male; F, female.
Musicians
88
79
74
Studio
mngrs
M F*
90
89
84
87
84
77
For a broadcast or to make a recording, the balancer monitors at a
fairly loud listening level with full attention. How does this compare
with most of the audience? In one experiment, checks were made on
the maximum sound levels preferred by BBC studio managers,
musicians and members of the public. As shown in the table it was
found that studio managers preferred louder levels than the musicians, and very much louder than the public.
For classical music, the figures for the public correspond reasonably
well to what they would hear at a distance in a concert hall. Musicians, normally much closer to the sound sources, might be expected
to choose higher levels. But the studio managers (sound balancers)
chose unrealistically high levels. One reason for such diversity is
that those professionally concerned with sound extract much more
information from what they hear, and the greater volume helps them
hear fainter wanted and unwanted sounds, checking technical quality
as well as the finer points of operations like fades and mixes.
Musicians, however, draw on their experience of instrumental quality
and musical form, and listen for performance, usually with less
concern for technical quality.
Those preferred maximum levels refer specifically to the case where
the listeners, like those concerned in creating the programme, do
want to listen attentively and have reasonable conditions for doing
that. But this is often not the case: for example, many people's early
evening listening may be while driving home or preparing food.
Where listeners' attention is limited and background noise may be
high, speech may sound too quiet unless it peaks 8 dB higher than
music. But if, when listening to a programme controlled in this way,
you decide that a particular item of music appeals to you and turn it
up, you will find that the announcement following it seems unduly
loud. As a general rule, listeners can do little to combat the levels as
transmitted.
Television sound in the home is often set at levels comparable
to those appropriate to background radio listening. This may be
satisfactory for some material, but if it is followed by a play with a
wide dynamic range the average level drops, and may not be corrected
by the viewer. The television sound supervisor (monitoring at high
level) may have set relative levels, of speech to effects or singer to
accompaniment, that are not appropriate to such quiet listening.
Alternatively, the viewer may raise the volume, and then subsequent
announcements and programmes (and especially commercials) are
much too loud: indeed, they seem all the louder because it is irritating to have to turn it down. For the broadcaster, all this requires a
fine judgement of compromises. Start this process by listening to
part of a rehearsal at realistic domestic levels on small loudspeakers.
Linking together speech items of varying intelligibility presents a
special problem, and particularly in the case where an item containing noise and distortion has to be matched to studio quality. Here,
the least unsatisfactory solution is found if the noisy item is matched
Volume and dynamics
189
for loudness at the beginning and end, and lifted for intelligibility in
between: the item might be slightly faded in and out.
Sadly, some readers will find that these suggestions about relative
and maximum volumes are impractical in today's commercial
environment, in which an overriding demand is for promotional
material to be as loud as possible, however much this may annoy
the audience. This inevitably distorts any ideal based on what the
listener would prefer, and leads to a general flattening of levels.
It also means that any attempts to apply aesthetically satisfying
standards must be rather more subtle than we might wish. Try not
to lose them completely!
In more practical terms, there is a further way in which listening
conditions may vary. At the one extreme, there are the noisy listening
conditions (in-car radio, for example) that demand high intelligibility
and a fully modulated broadcast signal: this contrasts sharply with the
quiet listening-room conditions and attentive audience (perhaps not as
common as we might wish) that requires `high-fidelity' sound with
a wide dynamic range. The two sets of requirements are largely
incompatible, and may best be served by different radio stations.
Dynamic range and compression
[10.8] Dynamic range and the need
for compression
1, The dynamic range of music can fill
the entire range of human hearing.
The maximum ranges for some
media are: 2, Videotape. 3, Vinyl
record. 4, FM radio. 5, Analogue tape.
6, CD, DAT, etc.
The ear can accept a great range of volumes of sound. At 1 kHz
the threshold of pain is 110 dB or more above the threshold of
hearing ± that is, a sound that is just bearable may be over a hundred
thousand million times as powerful as one that is just audible.
Studios or microphones with noise limits that are 10±20 dB above
the ear's threshold level are good for a range of perhaps 100 dB or
even more. Analogue recordings are limited to a signal-to-noise ratio
of 50 or 70 dB, but digital recordings raise this to a maximum of
80±100 dB (which demands a noise reduction system). FM broadcasting, back down at 50 dB, was actually a huge advance on AM at
30 dB. Even that may be more than some audience preferences or
listening conditions warrant.
Plainly, some compression of the signal is usually necessary. For
this, the old concept of `stops' remains useful. `Taking it down a
stop', a reduction of 1.5±2 dB on the fader, amounts to a drop in
level that will be just perceptible to the operator, but which the
audience will not associate with a hand on the control. A script or
score can be used during rehearsal to note settings and other information that will be useful on the take. The standard procedure is
roughly as follows:
Set the gain at a suitable average level for the type of programme,
and adjust the loudspeakers to a comfortable volume. Judge divergences from the average level by listening to both relative loudness
and intelligibility (remember that at times these two factors are at
odds). Glance at the meter occasionally, mainly to watch for very
quiet or loud passages. After recording, the effects of any high peaks
190
Volume and dynamics
can be rechecked, to discover whether there is appreciable distortion.
For broadcasts, an occasional peak may be better than reducing the
effective service area of the transmitter by undermodulating a whole
programme; compressors or limiters deal with peaks big enough to
cause technical problems. On speech programmes also check the
meter occasionally in order to maintain the same average volume.
The ear is not reliable for this: it tires.
At the end of rehearsal of a programme with recorded inserts, it may
be worth checking the levels of the first few inserts again, to see if
one's judgement of the right level for them has changed. It often has.
Compression of music
To reduce the extreme dynamic range of classical music, we should
again use manual control ± and also ears (both ears, for stereo), eyes
(for meter and cue-sheet or perhaps a musical score and to observe
actions that may affect sound) and brain, to anticipate events.
[10.9] Manual control of volume
1, Highest permissible level.
2, Lowest acceptable level. This
method of compressing a heavy peak
retains the original dramatic effect but
ensures that the signal falls between
the maximum and minimum levels.
Steps of about 1.5±2 dB are made at
intervals of, say, 10 s.
Avoid overmodulation on music that has loud and quiet passages,
not by `riding the gain' but instead by careful preparation, gradually
pulling back on the gain control for half a minute or so before a peak
is due, with subsequently a slow return to the average level. Each
operation should consist of steps of about 2 dB. During rehearsal,
appropriate settings ± the start point, total compression and end ±
can be noted on a score. A similar procedure may be used to lift long
periods of low level. It may also be necessary to vary the level
repeatedly throughout a work, so that all the elements of light and
shade fall within the acceptable dynamic range.
Compression of this kind is most likely to be necessary when recording orchestral works but, depending on the nature of the music and the
closeness of the balance, control may also be required for smaller
combinations or groups, or even a solo instrument such as a piano.
There is one other way of dealing with the control of orchestral
music: simply leave it in the hands of the conductor. It has been
known for a conductor to have a meter to check during rehearsal and
recordings, but this is unusual. On the whole, it seems preferable that
a conductor should concentrate on the performance, bearing in mind
the limitations of the medium if that is possible, but leaving control
to the balancer.
However, music specifically designed for background listening
may have some measure of compression designed in to it by an
arranger; and also, of course, for that, smaller musical forces will be
employed.
The further the microphone is from sources, the easier it is to control
music: studio or concert hall acoustics smooth out peaks. The balancer who adopts a bright, `tight' balance has more work to do
without necessarily earning the thanks of listeners ± who might
prefer a well-blended sound for classical music.
Volume and dynamics
191
Control of music affects the reverberation just as much as the direct
sound. The effect may be disconcerting: the microphone apparently
moves closer, but its perspective remains the same. The two may be
controlled more effectively by using two microphones (pairs or
arrays), one close and one distant, and working between them on
changes of music volume. Alternatively, reverberation microphones
may point away from the orchestra: they are controlled independently
to achieve the best effect as the acoustic volume of the music varies.
Compression: speech and effects
In controlling speech the problems are different, as are the methods
employed for dealing with them. Studio interviews and discussions
on radio and television also benefit by careful control, particularly if
one voice is much louder than the others, or habitually starts off with
a bang and then trails away. Here, control is a matter of anticipation:
whenever the needle-bender's mouth opens, drop the fader back ±
sharply if in a pause, but more gently if someone else is speaking. A
slight dip in level is more acceptable than the risk of a sudden loud
voice. Laughter may be treated similarly if the general level is held
well up. But edge the fader up again soon after the laugh.
Careful control should not be noticeable: do not miss the first high
peak and only then haul back on the fader. No control at all is better
than bad control.
Do not discourage speakers from raising their voices, unless they
actually ask for advice; even then the best advice is `be natural'. A
natural but controlled programme is better than one in which voices
are never raised and the meter never has any bad moments.
A good programme depends on variety of pace and attack. It should
not be necessary to hold back overall level in order to accommodate
possible overmodulation. But if editing is expected it is better to deal
with the more violent fluctuations of level afterwards, or there may be
difficulty even in a studio where extraneous noise can be kept low.
When recordings are made on location and subsequent editing is
envisaged, it is better that a level should be set such that background
atmosphere can remain constant throughout a sequence. Sometimes,
gentle control can be exercised by slight movements of a hand-held
directional microphone.
For scripted drama something like 16 dB may be an acceptable range
for peak values of speech, the average level being about 8 dB below
the maximum permissible. Even so, scenes involving shouting have
to be held back, possibly making them seem distant, which in turn
may not match a television picture. In cases where close shouting
must be mixed in with more normal levels of speech, the actors
should hold back on volume and project their voices.
This upper limit makes even more difference on loud, percussive
sound effects such as a gunshot or a car crash. Even a sharply closed
192
Volume and dynamics
door is too loud. It is therefore fortunate that radio and television
drama have established a convention in which such effects are
suggested more by character than by volume. Peaks must be evened
out and the overall volume held back. This explains why unlikelylooking methods for making sound effects sometimes produce
surprisingly good results.
Compressors and limiters
[10.10] Compression ratios
A, Linear operation: input and output
levels are the same. B, Compression
threshold. C, Various compression
ratios reduce the output level.
D, Extreme compression: here it acts
as a limiter, holding the maximum
output volume close to the threshold
level, but still permitting a small
proportion of any high peak to pass.
The full use of any recording or transmission medium requires that
some passages are close to the maximum permissible level, and that
they may momentarily or unpredictably exceed it. In such cases, the
danger of analogue sound distortion (or, far worse, digital clipping)
is replaced by a controlled (and usually less noticeable) distortion of
dynamics. After most corrections have been made manually, the few
that remain are `cleaned up' automatically by a compressor or its
more extreme form, a limiter. In popular music, these can also be
used to raise the level of individual channels in a mix, or to make the
whole recording denser or `hotter'. Essentially, a compressor works
as outlined below.
Below a predetermined level (the threshold or onset point) the
volume of a signal is unchanged. Above this threshold the additional
volume is reduced in a set proportion, e.g. 2:1, 3:1 or 5:1. For
example, if a threshold is set at 8 dB below the maximum and 2:1
compression is selected, signals that previously overmodulated by
8 dB would now only just reach full modulation. Similarly if 5:1 had
been chosen, signals that previously would have over-peaked by
32 dB now only just reach the maximum. In practice this means that
the overall level can be raised by 8 dB in the first case or 32 dB in
the second. Compression allows relatively quiet components of a
complex sound to make a much bigger contribution than would
otherwise be possible.
[10.11] Lifting low levels with a
compressor
1, Original signal. 2 and 3, Lower and
upper limits of required dynamic
range. With the compressor out of
circuit, only a small proportion of the
input goes into this range. 4, With the
signal lifted, increasing degrees of
compression allow more to be
accommodated. 5, The effect of
a limiter. The greater the
compression, the less the effect of
overmodulation (above 3).
If the compression ratio is made large, say 20:1, and the threshold is
raised close to the maximum permissible, the compressor acts as a
limiter. It can then be used either to hold back unpredictable high
peaks or to lift the signals from some lower level into the usable
working range of the equipment. Note, however, that high-level
signals may be excessively compressed and that background noise
will be lifted as well. Reduction of noise in gaps in useful material
(noise `gating') is described at the end of this chapter.
The effect of a 2:1 compressor with a threshold at 8 dB below peak
maximum is to compress only the top 16 dB of signals while leaving
those at a lower level to drop away at a 1:1 ratio. An interesting
variation on this would be to place two limiters together, working on
a signal in parallel but with the threshold set at different levels. If
one is set for 2 dB below 100% and the other for (say) 24 dB below,
but raised in level, this introduces a variable rate of compression that
is greater at the lower end of the range than near the top, where
Volume and dynamics
193
something approaching the normal dynamic range is used until the
upper limiter takes over. Another option is for the lower limiter to
have a weighting network to arrange that quiet high- or low-frequency
signals are lifted more than those whose power is mostly in the
middle (500±4000 Hz) of the audio range.
[10.12] Compression by two limiters
1, Effect of a low-level limiter.
2, Effect of a high-level limiter.
When the two signals are combined,
the compression characteristic is
different from that of a single-stage
compressor, which would give a
straight line from 3 to 4 and beyond.
Expanders operate as compressors in reverse, allowing further
fine control of dynamics. Used on music, they thin the sound and
selectively emphasize intermittent peaks. More often they are set to
operate only below a selected threshold, so that useful sound above
that level is unchanged but quieter sound in the gaps, which may
include unwanted noise, is reduced. Expanders can also be used for
more technical reasons, for example as part of the compander of a
radio microphone, where the signal may be compressed before its
local transmission, then expanded in the receiver to restore the
original dynamics.
In large consoles (especially those used for popular music), a range
of these controls is included in a dynamics module in line with each
channel fader (or, in `assignable' consoles, are available for
programming it). In simpler mixers they may be offered only on
output and, if needed for individual (usually mono) channels, are
inserted into the signal routing.
Compression using digital circuitry is very versatile. One device has
controls that can be set for expansion, compression or dynamic
inversion. The expansion ratio starts at 1:10 then proceeds by steps
to 1:1, after which it becomes normal compression. The scale continues to infinite compression (with a ratio of 1:0, a `perfect' limiter)
and then the ratio becomes negative, so that quiet sounds are made
louder and vice versa. This carries the compressor principle into new
realms, as an instrument of creative sound treatment and synthesis.
Compressors and limiters are also available as digital `plug-ins'.
Some programmes offer versatile high-quality effects, while others
use the versatility to recreate the peculiar characteristics of early
valve (vacuum tube) designs that were used in old pop music
recordings.
[10.13] Simple compressor/limiter
1 and 2, Threshold level control
(ganged). 3, Sample signal fed from
the main programme chain.
4, Feedback side chain with
compression ratio and recovery time
controls. 5, Control element operating
on the main programme chain.
6, Meter showing the degree of
attenuation. Alternatively, if a small
delay is introduced into the main
chain, a side chain taken off earlier
can intercept peaks in the delayed
signal before the first transient
arrives: it operates in feedforward
mode. Although this can be made to
work on an analogue signal, it is
easier when the signal is digital.
Compressor design and use
Compressors and limiters work by sampling the signal at a particular
point in the chain, and if it is above a designated level, deriving a
control signal that is fed back through a side chain to reduce the
overall volume in the main programme chain. It is usually arranged
that this control signal operates with a fast attack and slow decay
(also called release or recovery) ± though both `fast' and `slow' may
have a wide range of actual durations, with some controlled by the
balancer. A typical (compressor) attack may vary from 0.1 to 30 ms,
and the decay may take from 0.1 to 3 s. Other controls include
threshold and compression ratio, and there may also be a meter to
show the degree of attenuation.
194
Volume and dynamics
In popular music, the ideal attack and release times depend on the
instrument. For staccato sounds such as drums they can be set to
operate quickly; for sustained tones like those of a guitar they need
to be much longer, to avoid `pumping' effects. Experiment to find
the best settings for each instrument. For compression, try 2:1 or,
say, 4:1 on individual microphone channels, allowing their contributions to the mix to be boosted. It can also be applied to groups, or to
the master control to make the final mix denser and more powerful.
But beware: even for popular music, use automatic compression
with caution. What can add life can also take life away.
In some compressors the attack time can be reduced to a few tens of
microseconds. However, such a very sharp change of level is liable
to introduce an audible click or, short of that, to modify the transient
of a musical sound enough to change its character. The alternative is
to delay the whole signal for about half a millisecond, then make a
gradual change of level over the same period, in advance of any
sudden heavy peak. The transient problem is largely eliminated, but
any audible background still changes level abruptly, which is likely
to be noticeable and therefore objectionable. The quality of its
unpleasantness is also affected by the rate of decay of the control
signal. Heavy limiting, or strong compression with a moderate
recovery time, again produces mechanical and ugly pumping effects.
Early limiters had gain-recovery or control-signal decay rates of the
order of seconds: this allowed a reasonable time for the overload
condition to pass so that it would not be pumped by too rapid a train
of high-level signals. The more sophisticated limiters (described
above) have a selection of decay times available, typically 100,
200, 400 and 800 ms and 1.6 and 3.2 s. The fastest of these can hold
isolated brief excessive peaks without having a severe effect on the
background, but they are not so satisfactory for a rapid series of short
peaks: any background is given an unpleasant vibrato-like flutter.
[10.14] Compressor/limiter and
dynamics modules
Examples of in-line channel controls.
Left: Mono module. 1, Limiter
threshold and release (decay time)
settings and `gate' key.
2, Compressor threshold,
compression ratio, and attack and
release settings. 3, Meter showing
gain reduction due to compressor or
limiter. 4, `Make-up gain' amplifier
setting. 5, Switches for operation as
compressor, limiter, or under external
control. Right, Stereo module.
6, Expander controls. 7, Compressor
controls. Compressor/expander
modules may also be available for fine
control of popular music dynamics.
If there are only occasional peaks, and they are not too big, say not
more than 3 dB or so, the recovery time is probably not critical:
something like half a second may be adopted. The exact choice of
recovery time becomes important only at high levels of compression.
An alternative is to allow automatic variation between about 30 ms
and 10 s, varying with the level and duration of the overload signal.
`Gating' background noise
A major problem with limiters and compressors working on high gain
is that background noise is lifted too, unless something is done to stop
that. In particular, when there is no foreground signal to regulate the
overall level, that is, during a pause, the noise is liable to rise by the
full permitted gain. Such noise will include ventilation hum and studio
clocks, movement and other personal noises, traffic, electronic or line
noise, and so on. Even where such sounds are not raised to an
objectionable level, certain consonants at the start of subsequent
Volume and dynamics
195
speech may not be powerful enough to trigger the compressor. The
most unpleasant result is excessive sibilance at the start of some
words, though this can be avoided by high-frequency pre-emphasis
in the input to the limiter: this is sometimes called `de-essing'. It can
also be used to reduce sibilance on voices that naturally have too
much, or where a microphone with an inappropriate response has
added it.
[10.15] Noise gate
1, Range of maximum expected input
level. 2, Maximum output level.
3, Lowest level of useful input.
4, Lower end of desired dynamic
range. 5, Gating level: if level of input
signal drops below this, the output
falls from level A to level B, so that
noise in gaps between useful material
is reduced.
One device that can be used to discriminate against noise is a noise
gate, a specialized use of the expander in a dynamics module. In a
gate, when the sound level falls below another, lower threshold
setting, the gain is allowed to fall to a parallel lower input/output
characteristic: this might, for example, be 8 dB below that which
would apply without a gate. At this setting, a device that could not
previously have been used for more than 8 dB compression on a
particular signal without the background noise becoming excessively loud in pauses could now be used for 16 dB compression.
Noise gates are often made to work very quickly, so that even in tiny
pauses between the syllables of a word noise is sharply reduced. For
this to be successful, the settings must be chosen with care: the gate
must be below all significant elements of speech, but above all the
underlying noise. If these conditions cannot both be met, the result
can be unpleasant, with either speech or the noise wobbling up and
down on the edge of the gate. In any case, either extraneous staccato
noises at a higher volume or a rise in the background to an unexpectedly high level will fool the gate, as also (in the opposite sense)
would a very quiet passage at the start of a piece of music.
In the `automatic level control' available on some high-quality portable recorders used for professional radio and film location work, yet
another type of limiter is used: one that avoids most of the rise and fall
of background noise that is heard with other limiters. A short delay is
introduced before the control signal begins to decay. The duration of
this delay can be adjusted to allow for different lengths of pause
between syllables or words as spoken by different people or even in
different languages, but is normally 3 s, after which the `memory' is
erased and a new level of adjustment selected. This same recorder has
a separate circuit for the control of very brief transients, which do not
therefore affect the basic level. Obviously, such a device still has
limitations: there would undoubtedly be many cases where a human
operator would adjust the level, for example on a change of voice, so
it will probably be used only where it is inconvenient to have an
operator. Like the fast-operating gate, the device can all too easily be
fooled by unexpected events.
Chapter 11
Filters and
equalization
The word `equalization' reflects its historical origin, from when its
main function was to restore inadvertent changes to frequency
response. These were among the biggest obstacles to be overcome
as audio engineering has developed. Microphones still often introduce marked changes, differentially responding to the frequencies
present. Acoustics, too, may require correction.
For television drama, a range of different microphones may have to
be used to pick up a single voice as an actor moves around a set, and
the varied responses must be matched to some standard, perhaps of a
main boom microphone, or the transitions will be obtrusive. This
demands a capacity for changing each microphone's frequency
response in detail. So, when linking a film documentary, does the
jump between a personal microphone on location and a studio
microphone for commentary. `Equalization' rediscovers its original
meaning when applied to the correction of unintended changes like
these.
Today, equalization (EQ) is used just as much for the creative
manipulation of frequency response. In general, it is pop music
balance that makes the greatest use of EQ, with bass and treble
emphasized or rolled away, and selected bands of mid-range frequencies lifted to lend presence to particular components of the mix.
If frequency bands containing the characteristic elements of musical
quality of each instrument are raised, the result is a kind of musical
caricature or cartoon. In fact, that is precisely the object intended ±
and when it succeeds, the selected essentials do, indeed, seem to
press themselves upon the listener. Note that effects of this type may
also involve the use of compressors, as described in Chapter 10.
At its best, selective EQ, sometimes allied with compression, allows
contrasting sounds to blend together better, without loss of individual
Filters and equalization
197
character. At its worst, everything gets pushed up and the result is no
better than simply raising the main gain control ± which is pointless.
[11.1] Simple bass and treble
control
The effect of some domestic `tone'
controls.
In this chapter we are concerned with equalization, not just as a means
of correction, but also as a creative technique. Many examples of the
use of EQ in music have been given in Chapter 8, where EQ was
included as a component of music balance. Here many of the examples
are of the effects of EQ on speech, and particularly its use in drama.
Filters
The simplest kind of filter ± a passive filter using one resistor and
one capacitor ± cuts just the bass or top, rolling off at 6 dB/octave
above or below a given frequency. At the nominal cut-off frequency
the signal is 3 dB down. A band-pass filter has two filters and the
designated band lies between points that are 3 dB down at each end.
Steeper roll-off at, say, 12 or 24 dB/octave requires more complicated networks. Active filters have amplifiers built in, to lift peaks to
higher levels or make up for losses. Boosting very low frequencies,
say below 100 Hz (below 60 Hz for subwoofers in larger auditoria),
adds the power sometimes felt in surround sound. For some instruments, boosting lower frequencies can add warmth. Boosting very
high frequencies, above 5 kHz, can add brilliance ± or, more likely,
harshness.
[11.2] Band-pass filter
This is a combination of high- and
low-pass filters. The band is defined
as the range out to 3 dB down at
either end.
Variations on these effects require greater complexity. An active
midlift filter with a broad or narrow peak is used for presence and is
a vital part of pop music balance. The middle of the audio range may
be taken to be about 256±440 Hz, but presence, more often applied a
few octaves above that, acts on the harmonics of the notes produced,
strengthening and extending them. As different notes are played by
an instrument, the presence frequency remains constant, so it
behaves rather like a physical part of that instrument, creating a formant
± or, more likely, enhancing a formant already present that helps
define its character. Midlift in the 2±5 kHz range enhances the clarity
or definition of many instruments such as a snare drum or woodwind.
Vocal formants in the middle of the audio range shift continuously in
the process of speech, but the higher formants are caused by resonance in cavities of less flexible or fixed size. For presence on vocals
it is usually the upper range that is enhanced, perhaps by 6 dB
somewhere in the 2±5 kHz range, and this can also add intelligibility.
[11.3] Simple `presence' boosting
The effect of an early two-stage,
four-band midlift system.
Note, by the way, that the recommendations here and elsewhere in
this book are necessarily imprecise, since the final effect must
always be selected by ear in the light of particular qualities of each
instrument, the relationship of one to another, and the overall melodic or harmonic effect desired. Also note that EQ can also all too
easily be overdone.
Early presence devices had a limited range of possibilities, offering
perhaps four or five nominal frequencies separated by intervals
of rather less than an octave, together with several degrees of lift,
198
Filters and equalization
[11.4] Some commonly used EQ
symbols
switched at increments of 2 or 3 dB. In more complex designs the
range is more flexible in both frequency selection and volume, and
will also allow for a dip in place of a peak: by analogy with
`presence' the effect would be `absence', and this too may be used
in pop music mixes. Shelving, flattening out beyond a high or low
filter, can be used to hide problem areas, such as boominess,
scratchiness or sibilance. It can also help to improve separation,
reducing an unwanted range picked up on the wrong channel. For
this kind of filter, the signal may be diverted into a `shift register', in
which there is a delay of the order of microseconds that is different
for every frequency present: the longer the wavelength, the greater
the delay. When recombined in different strengths with the original,
the changing phase differences produce the desired `shelf' effect.
A variable in more flexible systems is the narrowness (known as the
`Q') of the peak or dip. In most consoles today, a group of controls
for bass and treble cut (or shelving) and several degrees of presence
is provided on all of the individual channels, mono and stereo, and
also in group fader channels. If `Q' is marked on a control or quoted
in a specification, this is calculated as the centre frequency divided
by the bandwidth (between frequencies where it is 3 dB down).
A Q of 1 covers more than an octave; a Q of 10 picks out individual
tones.
On a complex analogue console, each mono channel may have as
many as a dozen rotary controls and a variety of other switches. The
high- and low-pass filters remain relatively simple, with (in one
example) an 18 dB/octave cut at a choice of ten turnover points from
3 kHz up and 12 dB/octave cut at another ten from 350 Hz down. The
equalizers operate in bands of 1.5±10 kHz, 0.6±7 kHz, 0.2±2.5 kHz
and 30±450 Hz, each with a range of peak and shelf settings. In
addition, two in the middle have variable Q. There is an overload
indicator light, monitoring for this at three points in the EQ module.
The stereo modules have fewer controls, but each is available over
a wider range; and overload is indicated separately in the
two channels.
If the signal passes through a filter network with everything set to
zero, an analogue-controlled setting can never be quite as smooth as
the original, so it is usual to include a bypass switch which, in effect,
switches the whole module on and off. It is also convenient to be
able to switch any filter in or out, so that direct and corrected sound
can be quickly compared.
[11.5] Equalization modules
In-line channel controls on a console
designed for mixing popular music.
Left: Mono channel. 1, High- and
low-pass filters. 2, HF equalization.
3 and 4, High and low mid-band
equalization. 5, Variable Q
(bandwidth). 6, LF equalization.
7, Routing switches. Right: Stereo
channel. 8±10, HF, mid-band and LF
equalization. All equalizers have a
wide choice of centre frequencies
and boost or dip levels.
In some computer-controlled systems, the settings may be selected
and assigned from a central module that is separate from the individual channels, but stored in memory to operate on them.
`Notch' filters (with very high but usually undefined Q) are used to
eliminate sharply defined whines or whistles such as the television
line frequency at about 15 kHz. A dedicated hum eliminator takes
out mains hum at 50 or 60 Hz, together with its main overtones. Such
specialized filters are not needed on every channel, so are patched
in at an insert point when required. Increasingly, digital signal
processors are used; these have a wider and more flexible range of
Filters and equalization
199
functions and can be tuned far more precisely than their analogue
predecessors.
When used for EQ, digital signal processing programs often emulate
the naturally smooth curves of analogue EQ (and often of existing
analogue EQ characteristics), but with the potential for greater
versatility.
Graphic equalizers
[11.6] Graphic filter
In one design the slide faders are at
one-third-octave intervals, and control
the gain or attenuation in a band of
about that width. In principle, the
overall effect should match the line
shown on the face of the module, but
this may not be achieved in practice.
The result should be judged by ear.
Even in combination with bass and treble filters there is a limit to the
flexibility of equalization based on individual presence controls. In a
television studio or when filming on location, the degree of mismatch between microphones can be more complex. One approach
has been to take the principle of midlift further and split the whole
spectrum into bands, typically about 30, each of one-third of an
octave, and apply correction separately to each. If all controls are set
at zero, to make a horizontal line across the face of the panel, the
response should also be level throughout the audio range; in practice, the actual line may be distinctly bumpy and there is a surprising
amount of difference in performance between units from different
manufacturers. In any case, the unit should be switched out of circuit
when not in use.
If individual sliders are raised or lowered, their associated circuits
form peaks or troughs that overlap, each blending into the next. The
selection of the desired effect is made by setting the slides for
the various component bands to form what appears to be a graph of
the new response. In fact, the actual response achieved may be different:
it depends on the design of the filter and the size of the fluctuations
set. The true arbiter ± once again ± must be subjective judgement.
Equalizers of this kind are plugged (patched) into the channels
where they are needed. So, too, would be the digital alternative.
One `plug-in' stores `models' of many present-day and older microphones, together with the effects of windscreens, polar response,
proximity effects and switchable filters on their frequency response,
so that by entering source and target microphones, one can be
`converted' to the other.
[11.7] Audio response control
Ideal effect (A) of a single fader and
(B) two adjacent faders set at
opposite extremes. In practice, sharp
jumps in the response can only be
achieved at the expense of a more
erratic response when the sliders are
set for a smooth curve. Apparently
similar units may differ in their
capacity to compensate for this.
Graphic filters have been used in film dubbing theatres, on television
control consoles and more generally for any creative sound application. When speech is put through one, it is possible to make severe
changes, for example to introduce several dips of 30 dB without
much reducing intelligibility. If this is intended not just to restore
the deficiencies of a balance made in adverse conditions, but instead
to achieve a positive and marked dramatic effect, the imposed
distortion has to be drastic. However, the selective emphasis of
particular frequencies is noticeable where a dip might not be. The
brain reconstructs defective harmonic structures, perceiving sounds
as experience says they should be, but finds it difficult to disregard
what should not be there, but plainly is. Given that peaks are more
200
Filters and equalization
noticeable than dips, a useful strategy for locating and reducing
distortion or noise that affects a limited but unknown frequency
range is to lift the faders successively, until the problem sounds
worse, then reduce its setting to below the median.
Telephone effects
Telephone circuits often have a bandwidth of some 3000 Hz with a
lower cut-off at about 300 Hz, but this should not necessarily be
copied when simulating telephone quality. Settings for bass and
treble filters should, as always, be made by ear. How much is cut
may vary even between different simulated telephone effects in the
same presentation, just as, in life, the amount of the lower frequencies can change with how closely a telephone is held to the ear.
[11.8] Simple telephone effects unit
The example has four settings for
bass-cut (B1±4) and five for treble-cut
(T1±5).
A band-pass filter is used for telephone effects, but there is no set
range for this. It depends on differences between male and female
voices, and between different examples of each. If the fundamental
of a male voice is at 110 Hz, then bass-cut at 220 Hz will give a
small but noticeable change of quality. For a female voice an octave
higher, 440 Hz bass-cut is more suitable. In practice, 440 and 880 Hz
may be better still: the loss of bass is stressed for dramatic effect.
The top-cut will be closer to that of the telephone circuit. A microphone used like this may be called a filter microphone or perhaps
a distort.
In a full-scale studio telephone conversation set-up it is usual to have
two microphones, one normal and the other with the filter. In radio,
these two can be placed in different parts of the same studio, but not
separated so much that the actors cannot hear each other direct. This
arrangement avoids them having to wear headphones.
It is important to avoid acoustic spill, which in this case means pickup of the supposedly filtered voice on the normal microphone:
otherwise the effect of telephone quality will be lost. If the `near'
microphone is directional, and it is placed with its dead side toward
the `far' voice, the pick-up of direct sound is effectively prevented.
Discrimination against reflected sound is more difficult, so it is
better to use a dead part of the studio for the filter microphone.
The two microphones are mixed at the control console and the
relative levels are judged by ear. With this sort of filter effect, meters
are even less use than usual in matching the voices, as the narrowband input has less bass, and therefore less power, for similar
loudness. At normal listening levels there is little difference in
intelligibility between full-range speech and the narrow band used
to simulate telephone quality, so a fairly big reduction in the loudness of the filtered speech may be tolerated in the interests of
creating a feeling of actuality.
In a television studio the two microphones should be placed well
apart. Then, as the picture cuts from one end to the other, the normal
filter arrangement can be reversed without spill either way. If the
Filters and equalization
201
[11.9] Telephone effects switching
unit for television
1, Pre-selector switches. 2, Switch
linked to video-cut buttons. Any
sound routed through this is
automatically switched in and out of
the filter condition when cutting
between pictures. 3, Bass and treble
filters. In this example, microphone
A has full sound only while B is
filtered and vice versa, and
microphone C remains the same
whatever the picture.
selection is linked to the vision cutting buttons the sound and picture
changes can be synchronized. But it is best not to cut in the middle
of a word as this sounds unpleasant and draws attention to the
technique.
Note that as distorted sounds must be held at a lower level than
`clean' sound, care must be taken not only in balance, but also before
that, in studio layout. If the two performers cannot be acoustically
completely separated, because both must appear in vision and must
therefore be in the same studio, it should be arranged for the two
settings to be as far apart as possible. If they are not, coloration can
affect both voices, as distant, reverberant but otherwise undistorted
sound is superimposed on the telephone quality speech. Also, if one
performer speaks louder than the other, the loud voice is likely to be
heard on the quiet speaker's microphone. It may help for the actors
to face each other in the studio layout, so that directional microphones point away from each other. But note that the use of wide
shots makes it worse, because in those the microphones are farther
away, so it more likely that the distant voice will be heard as the
channel gain is raised.
Background effects added to a telephone conversation can make a big
difference to the way voices are matched. The degree of filter and its
volume relative to the clean speech must depend on both the loudness
and the quality of the effects. In particular, `toppy' background sound
tends to blot out quiet speech, but if the backing is woolly or boomy
the shriller telephone effect cuts through it. A complicated scene may
require varying intelligibility, as for example in this radio script:
EFFECTS
GUEST
CHARLIE
GUEST
CHARLIE
GUEST
CHARLIE
GUEST
CHARLIE
(Loud party chatter background; phone dialling close, ringing heard on
line: it stops.)
Hey, is that you, Charlie?
(On filter, barely audible against party chatter.) Yes. This is Charlie.
Say, this is a great party, why don't you come on over?
What are you at, a football game or something?
What am I what?
I'm in bed.
Hang on, I'll shut this door. (Door shuts; chatter down.) S'better! Now
wha's all that about a football game? It's three in the morning.
(Now more distinct.) Yeah. Three in the morning. I'm in bed.
202
Filters and equalization
Two different mixes are required for this, and the balance and pitch
of the near voice must change on the cue `door shuts'. In addition,
the backing should change in quality too: heard through a real door,
such effects would become muffled and muddy, as well as being
reduced in volume. Some degree of top-cut might be introduced at
that point.
There are other dramatic uses for filters besides telephone quality:
for example, to simulate station or airport announcements (adding a
touch of echo) or intercom (at the same time simulating a poor radio
link by playing in a crackling, hissing background). A film, television or radio play may include the effect of a radio announcement
(or tape replay) for which a useful convention is that the reproducer
has little better than telephone quality. This again is a caricature in
which one type of distortion, reproduction of sound through a single
loudspeaker, is represented by another, limitation of frequency range.
Chapter 12
Reverberation and
delay effects
This could be a very short chapter. You want reverberation? Just buy
an artificial reverberation (AR) device, put it in a channel and power
it up. Do you want really long reverb? Turn the knob to, say,
`5.5 seconds'. Maybe also scan through a seemingly endless list of
acoustic environments to reach `cathedral'. Plug it across the channel you want to reverberate ± and you are in business. You are
talking, walking or playing your trumpet ± in a space like a cathedral.
And, just like it says on the wrapper, the quality is amazingly
realistic. The same goes for other `delay' effects. Just plug in the
`reverb' device, check the instructions and off you go!
Digital AR and delay-related effects are the ultimate black box
success story ± particularly when compared with some of the
alternatives.
The longer version of this chapter does just that. It explains what AR
tries to emulate and describes some of the earlier more-or-less
successful attempts to solve a seemingly intractable problem. In
the past, various forms of artificial reverberation have been used,
including echo chambers, plates and springs. These all differed in
quality from one another, and from the real thing. Interestingly,
despite daunting limitations, some really do work well, and others
that you might find gathering dust in a storeroom have on occasion
been disinterred and brought back to life to show off their strange
`retro' character. Rather more easily you could run down to `plate',
just one of a hundred descriptions offered on the AR list, retained for
the benefit of those who recall and can still find a use for its distinctive
response. But now that artificial reverberation can be constructed to
mimic the characteristics of natural acoustics so much more accurately, most users, for most purposes, will now prefer to head for a
rather more alluring environment ± if only as their starting point.
204
Reverberation and delay effects
This chapter covers the techniques for using AR (or even real
reverberation) and also a related group of delay effects, in a wide
range of applications. Artificial reverberation, too, can be an effect
in its own right, but is more often used when more (and perhaps also
better) reverberation is wanted than the built-in acoustics of a studio
can supply. For music it is used to enrich a dry environment, to
rebalance the contribution of spotting microphones, or in a close
multimicrophone balance to supply a vital component of the mix
that would otherwise be missing.
[12.1] Echo and reverberation
1, An echo is a discrete repetition of a
sound after a delay of 20 ms or more.
2, Delays shorter than 20 ms (which
are associated with the dimensions of
all but the largest halls) are perceived
as components of the reverberation.
The term `artificial reverberation' has at times been shortened to the
more convenient `echo', but that is something of a misnomer for a
studio technique that extends reverberation without, it is to be hoped,
introducing distinct repetitions. A true echo may be heard only when
reflected sound is delayed by a twentieth of a second or more. (How
short the delay can be depends on how sharply defined the sound is.)
As separation is reduced, there comes a point when even a staccato
original and its repetition are no longer perceived as separate entities, but become a single extended sound, a composite that has
changed in character. There is, in fact, partial cancellation at some
frequencies.
Delay effects may fall into either category: they may be short, to
make use of the phase cancellations; or long, producing perceptibly
separate echoes, perhaps with further repetitions. They can also be
used simply as a time-shift device, for example, to move a closebalanced component farther back, so it does not stick out in front of
a widely balanced whole. In this case, delay will be combined with
AR. Delay-related effects also include pitch and/or duration changers.
AR feeds in the console
[12.2] Echo mixtures
1, Original signal. 2, Reverberation.
With separate control of direct sound
and artificial reverberation, a wide
range of effects is possible.
A, Reverberation dominates the mix.
B, Reverberation tails gently behind.
Because AR and delay-related effects devices are nearly always
digital, when used with analogue consoles they need A/D conversion
before the process can begin, and D/A conversion to feed the result
back into the signal. This is all quite routine: the converters are part
of the `black box' that contains the device. But the box in question
generally lives outside the console and is plugged in to the channel
(or group of channels) as required. A great variety of these devices is
now available, offering a range of AR quality and an even greater
range of other delay-related effects that may come with the AR or
separately packaged. Inevitably, this kind of device is liable to be
more expensive than a purely digital program in a digital console (or
for the computer at the heart of a `virtual studio'). This is because
of the cost of all the hardware (the box) plus the additional A/D and
D/A stages.
In fact, the device is just a computer in disguise ± together with its
built-in sound card (which does the converting). AR and delay require
a processor and some random access memory (RAM). Digital samples of the data (the sound, now converted into pure information) are
fed to memory, then after a pause for as long as the sound would take
Reverberation and delay effects
205
to travel some given distance it is read out again, perhaps slightly
reduced. For a computer, this is a simple operation, although for
good AR, it can take a very large number of operations like that, in
many different distinctive combinations. In principle, any other delayrelated effect is much simpler: all it takes is a separate set of instructions (the program) and a channel and some switches to feed it out.
[12.3] Flexible `echo send'
1, Channel fader. 2, Channel output.
3, Pre- or post-fader switch. 4, AR
send fader. 5, Bus-bar allocated to
AR. 6, Feed from console to AR
device. On many consoles, the feed
to AR passes through auxiliary `send'
circuits that may also be used for
other purposes. In the simplest case
these are taken from a point after the
channel fader, so that when a channel
is faded out the reverberation goes
with it. In a complex balance, several
AR devices may be used, each fed
from a separate bus-bar.
Most consoles have provision to split the feed from each source
channel fader, diverting part to a range of `auxiliaries', one or more
of which can be sent to the AR and effects devices located outside of
the console. In many consoles, auxiliaries can be fed from a choice
of points, either before or after the fader. For AR, the signal should
pass through the channel fader first. (If it were to go to the auxiliary
feed before reaching the fader, the AR return level might have to be
adjusted every time the source channel was faded up or down.)
If the whole of the sound from a particular single source is to be
passed through a delay device, this may be cross-plugged (patched)
directly into the channel, and in this case it can be before the fader.
For stereo echo, A and B signals are often combined before being
fed in, but A and B reverberation is extracted separately. This also
means that mono sources taken as part of a stereo mix automatically
have the stereo reverberation that is essential to a realistic sound.
The reverberation from each device is returned to the console as a
line source (usually stereo) in a new channel, so it has its own fader
(or faders) and equalization.
Since the console and the AR devices are completely separate, there
is also freedom to use any other means of creating reverberation. It
does not even have to be artificial. It can be real. After all, real (at
least in principle) is the standard by which any alternative is judged.
Real reverberation
Natural reverberation begins with character-defining first reflections
from the roof and walls ± in the centre of a concert hall, after a
substantial delay ± followed at progressively shorter intervals by
many more, smaller products of multiple reflections until, vast in
number, they trail away below the threshold of hearing. This is the
process that most Western instruments and ensembles were designed
to exploit: it is an unwritten part of the music. When considering
alternatives to natural reverberation we must therefore ask how their
decay effects compare with the natural form and its many variations,
not all of which are benign. An alternative could magnify the
acoustic infelicities, or introduce new ones. At the other extreme,
it might eliminate all the natural rough edges, and sound not just
smooth but lifeless.
In the music room or concert hall, some frequencies are absorbed
more readily than others at each reflection; if after many reflections
some are sustained while others diminish, the result is coloration.
206
Reverberation and delay effects
Coloration occurs in both natural and artificial reverberation, but
note that the build-up and blending of natural reverberant sound
depends on a three-dimensional system of reflections and decay that
greatly enriches the quality at just the stage when coloration is likely
to be most severe. When the artificial reverberation differs from
natural acoustics, its use is justified if the new pattern of coloration
and density of decay products is still perceived as musical in quality
± and if its origins are not obtrusively apparent, as they were in some
of the older systems. When real reverberation is added from a source
that is remote from the original studio, this must also be judged by
the same standards. How good, how natural, does it really sound?
Historically, the first way of enhancing reverberation was to use a
separate room, often called an echo chamber. (Properly, it should be
called a reverberation room, of course, but that is a lot less snappy.)
This `chamber' certainly had the advantage of real acoustics,
although often in far too small a volume. But real acoustics are still
a real option, so this needs to be considered not just as a historical
curiosity, or even as a baseline for comparison, but as a practical
possibility.
[12.4] Echo chamber: size
A U-shaped room may be used to
increase the distance the sounds
must travel from loudspeaker to
microphone. Here the interior wall is
angled to reduce standing-wave
coloration.
A typical echo chamber has reflecting walls, perhaps with junk
littered about at random to break up the reflections and mop up
excess mid-range reverberation. If space is limited it might have a
wall down the middle, so that sound has to follow a U-shaped path in
order to delay the first reflection. The loudspeaker itself can be made
directional so that no direct sound is radiated towards the microphones, which in turn can be directional, discriminating against both
sound from the region of the loudspeaker and coloration due to
structural resonances (between parallel roof and floor, for example).
A humid atmosphere gives a strong high-frequency response; a dry
one absorbs top. An echo chamber linked to the outside atmosphere
therefore varies with the weather.
When a large hall is simulated, there should actually be a reduced
high-frequency response, corresponding to the long air-path. In
some chambers, the bass response may be too heavy; or alternatively
the original sound may already have sufficient bass reverberation, so
that none need be added. Equalization (EQ) and bass-cut controls in
the return channel allow the necessary adjustments. Note, incidentally, that if less bass is fed to the chamber, its loudspeaker can be
used more efficiently, but if bass is reduced for this reason it needs
subsequent equalization.
Because it takes up valuable space within a building, an echo
chamber may be rather small, perhaps 110 m3 (4000 ft3) or less. This
will have broadly spaced coloration, due to natural room resonances
in the lower middle frequency range. Like radio studios, it needs to
be isolated from structure-borne noise, and can be put out of commission by building or engineering works unless noisy work is
stopped while it is in use.
Another disadvantage is that, once laid out, its reverberation time is
fixed. If this is about 2 seconds it may be suitable for music but little
use for speech. If, however, only a touch of echo of this duration is
Reverberation and delay effects
207
used it is noticeable only on the louder sounds, a situation that
corresponds well with what we sometimes hear in real life. In this
example an overlong reverberation time is apparently reduced.
[12.5] AR: the effect of reduced level
1, Loud direct sound. 2, Audible
reverberation. 3, Quiet direct sound.
4, Reverberation is now below
threshold of hearing (5).
For music the (fixed) reverberation time may be about right, but with
too early a first reflection (because the shortest path length between
loudspeaker and microphone is far less than in a bigger hall). To
restore both clarity of attack and a sense of spaciousness, a delay
should be introduced into the echo return feed.
Far better results have been achieved by using a spare empty studio
or full-sized hall as the echo chamber. This is rarely practicable, but
if it is, it is well worth considering for works that might really
benefit from the added spaciousness. The real luxury is to start out
afresh and lay out a new, dedicated space for this purpose. It is then
also possible to fit movable doors and absorbers to make the acoustics variable.
Analogue AR
The first reasonably successful AR device ± more versatile than
most echo chambers because reverberation time could be varied ±
was the reverberation plate. In principle, this was rather like the
sheet of metal once used in theatres to create a thunder effect, except
that, instead of being shaken by hand and delivering energy directly
to the air, it had transducers. One vibrated the plate, working much
like a moving-coil loudspeaker; two others (for stereo) acted as
contact microphones, to pick up the vibrations. The plate was 2 m2
of tinned steel sheet stretched tight inside a steel frame. Unlike the
echo chamber, its natural resonances at low frequencies did not thin
out to sharp peaks in the bass, but were spread fairly evenly throughout the audio range.
[12.6] Reverberation plate
1, Metal sheet. 2, Steel mounting
frame. 3, Damping plate. 4, Pivots
on damping plate used to change
the spacing and so change the
reverberation time. D, Drive unit.
M, Contact transducer. The unit shown
is mono; stereo has two contact
transducers, asymmetrically placed.
The (asymmetric) layout of the transducers, together with the frequency characteristics of the plate, its variable damping and the
response of the transducers and their amplifiers, were all arranged
to give a maximum duration of reverberation at mid-frequencies. It
had some roll-off in the bass but more at higher frequencies (15 dB
at 10 kHz) to simulate the high-frequency absorption of air in a room
of moderate size. With the broadest separation of damping and plate,
it had reverberation of 5.3 s at 500 Hz, dropping to 1.5 s at 10 kHz, as
in a cathedral. With increased damping this could be reduced to
under half a second, although in practice it was better when used for
reverberation times suited to music rather than for speech. A short
direct path between the transducers was unavoidable so, again, to
simulate a large hall, delay was required.
The 2-D structure of the plate allowed the wave to radiate outward
and break up, producing reflections that arrived in increasing numbers as the reverberation decayed. This was a fairly `natural' effect,
although distinctively different from true 3-D, especially on string
tone.
208
Reverberation and delay effects
A third, and rather more weird, medium for artificial reverberation
employed what were usually described as `springs'. They really
looked like springs, but worked quite differently: the acoustic input
was applied torsionally to one end of the helix, and sensed at the
other. The metal was etched and notched along its length, to create
internal reflections, each of which fathered its own family of subsequent echoes, and so on. Early, inexpensive systems of springs
were developed for use in electric organs. With improved engineering, they produced a quality of reverberation comparable with
chamber and plate. With reverberation times of 2.0±4.5 s they were
not suited to speech, but they were better than their early rivals for
string tone.
Plates and springs may still be encountered and used for `retro'
special effects, but are hardly worth the effort or expense of maintaining them. Echo chambers, too, are pretty much an expensive,
acquired taste. Digital techniques now cost less and are far
more versatile.
Digital AR
A typical good digital AR device has a very wide range of reverberation characteristics, many of which are pre-set, but with an
additional capacity for customized programs, which can be saved
and reused. As with other advances in computer technology, this has
depended in part on vast increases in random access memory
(RAM), together with high-speed processing and improved software. Each AR program will have several stages. First, there is a
group of delays that are characteristic of some particular shape and
size. These are followed by a second generation of delays with
feedback loops set for gradual decay, and the more stages that are
added, the richer and more flexible it becomes. There may be
separate programs for specific delay effects.
[12.7] Digital AR control unit
The deceptively compact controls for
one `digital signal processor' that
offers a wide variety of AR and other
effects. Features and controls
include: LED display showing a
descriptive name of the selected
program, its main characteristics and
the volume (including overload
warning); program or register keys;
display showing number of program
or register selected; function keys;
sliders and keys associated with the
currently selected control `page'.
Operation is not intuitively obvious:
new users should follow the
step-by-step instructions, guided
initially by the supposedly
`user-friendly' program names.
Few operators use more than a limited range of the facilities made
available by any computer system ± life is too short ± but it is worth
experimenting. The manuals and instructions are often so full of
jargon that they are likely to put off all but enthusiasts, but among
their wild swings between incomprehensibility and oversimplification, there are likely to be nuggets of indispensable information that
apply to the system in use and no other.
To learn to operate a relatively complex (and therefore versatile)
device, first run through a few of the existing programs in memory.
As well as displaying strings of numbers, these may have been
assigned `user-friendly' names, but it is best to take these purely as
a general indication of content and as a starting point in the search,
rather than expect a programmer's idea of a particular acoustic to
match your own. Check how to select from what is on offer, and also
how to store your own modified (or originated) acoustic programs.
The simplest strategy is to use only those already available, of which
Reverberation and delay effects
209
there is usually a range that is good enough for most purposes, but
you may soon want to start `tuning' them. A simple way of learning
to use the controls (press buttons, knobs and sliders) is to convert
one existing program until it matches another already in memory:
the match is confirmed by the series of numbers ± `parameters' ±
which define the acoustic effect. Then add the original and the
intervening stages to a short, dry recording in order to monitor the
changes.
The next stage is to alter the parameters to something completely
new. Here there can be an inhibitingly large range to choose from,
and you may suspect that some offer different routes to the same
objective. One device offers 22 parameters that can be altered to
modify reverberation, with another 13 to develop `effects', all within
a stereo image. The aim is to control each stage of the simulated
acoustic reinforcement of a sound, beginning with the `pre-delay'
(i.e. arrival time of the first reflection, in ms). Controls for the
character of early reflections may be described vaguely as its `pattern' (of coarseness and irregularity).
A suitable setting for the initial `pre-delay' not only creates necessary effects of distance and spatial volume, but also allows the
transients of the original sound to be heard clearly before the arrival
of any reflected sound, of which the initial volume ± its `attack' ±
may be another separately controlled parameter. Other controls may
be for relative levels of high and low frequencies. (`Room size' ±
area in m2 ± may also be offered.)
The most important single measure of an enhanced, natural-sounding acoustic is its reverberation time. With digital AR this can
usually be set for different values in several frequency ranges.
Depending on its period and style, classical music generally sounds
best in a hall with a mid-range decay of 1.8±2.5 s. However, low
frequencies may require longer, say 2.5±3.0 s, and it will be necessary to set a crossover frequency at perhaps 600±800 Hz. Treble
decay may need to be shorter, with a crossover at 6 kHz or above.
In addition, the length of decay in one or more bands might be made
to depend on the presence or absence of gaps in the music, so that for
vocal or instrumental clarity it is relatively short when the sound is
continuous, but longer (and so richer) in pauses. This is a modification that will be used cautiously: on lead vocals it helps to define
phrasing, but if it is also applied to supporting lines that have gaps in
different places, it may muddy and confuse the overall effect. In a
reversed or `stooped' version of this, a component can be given rich
reverberation as it runs continuously, but rapid decay to clear any
pauses in order to allow intervening figures to stand out in them.
Advances in computer technology have been such that extremes of
complexity (often including a wide range of other delay-related
effects) are now available for far less than the price of a plate ±
versions of which can in any case be mimicked by most AR programs, so that balancers who had learned to make constructive use
of its severe limitations need not feel deprived: they can have a
known (if simulated) reference point to move on from.
210
Reverberation and delay effects
Less complex devices can now, in principle, be provided at much
lower cost, but will still do far more and with better quality simulations than in the systems they have replaced. But while the choice of
AR and delay-related effects has grown so much, the microphone
techniques for joining them with live sound have remained much
the same.
Using AR on speech
[12.8] AR for perspective
A second directional microphone is
suspended above the main
microphone. If the upper microphone
has a stronger echo feed this adds
depth to crowd scenes. Here
bidirectional mono microphones are
shown; for stereo, try two crossed
pairs. This layout is for radio, but the
same principle can be used for
television.
Because echo chambers and early devices such as the plate were so
poor at providing short reverberation times, one of the greatest
improvements offered by the use of digital reverberation is found
in its application to speech ± usually for dramatic effect. The main
limitation is now due to the studio itself, and the performer's working distance. An actor who works close to the microphone and
whispers will, for the same setting of the echo fader, produce just
as much reverberation as when standing 3 m (10 ft) away and shouting. So compensate for perspective effects: as an actor moves in
toward the microphone, fade down the AR.
If there are other voices or sound effects that have to be picked up in
a different perspective at the same time, this requires a more complex set-up. One useful arrangement has a second directional microphone (or stereo pair) with a strong AR feed suspended above a main
microphone that has little or none. An actor who moves in close to
the lower microphone is no longer in direct line with the upper one.
In this way, perspectives can be emphasized without constant resort
to the AR fader. This is more likely to be of use in radio than in the
visual media, where there are restrictions on the ideal placing of
microphones, but the same principle (of direct and AR microphones)
can still be used.
Delay-related effects
A word of warning here: any of the effects described below can
easily be overused; for example, echo effects (typically with a delay
of about 35 ms) are often most effective when applied to a single
phrase of music, repeated only when that phrase recurs.
[12.9] Separate AR on two voices
If differing amounts of echo are
needed on two voices taking part in
a conversation, it helps if opposing
microphones and voices are placed
dead-side-on to each other. For
stereo, the two can be steered to
suitable positions within the image.
This layout might be adopted in a
radio studio; in a television studio
there is normally more space
available, making separation easier.
Some of the less natural effects of feedback and delay had already
been explored and developed by more laborious techniques long
before the present digital systems became available. For example,
a simple echo could be achieved just by allowing tape to travel from
the recording head to the replay on a tape recorder, though for any
flexibility, the machine needed a variable-speed drive. If the output
was mixed back into the input, still with a delay of 35 ms or more,
the echo circulated, dying away if the feedback was attenuated, or
becoming a howlround if it was amplified. This effect is sometimes
called spin, recirculation or slap echo, and has a mechanically
Reverberation and delay effects
211
repetitive quality. The effect varies with the length of the delay: at a
fifth of a second, for example, there is a marked flutter effect. All of
these, and many more, can now be selected from programs of digital
delay effects and control parameters.
For example, it is possible to start with a staccato effect such as a tap
on a glass and let it spin indefinitely by controlling the feedback
volume manually. Before the advent of digital technology, any slight
unevenness in the frequency response of the circuit was quickly
exposed: using tape, even a high-quality recorder soon revealed its
deficiencies and within half a minute the coloration characteristic
of the system completely swallowed the original sound. This effect
(or defect?) can now be eliminated.
It was also found that a signal delayed by about 15±35 ms and added
to the original could be used to double the apparent number of
instruments. For a while, `Automatic Double Tracking' (ADT) was
provided by a separate special digital device; now more often it is
just one of the many options on a digital signal processor (DSP).
Doubling (or tripling) works well with sounds that are sustained and
have a broad frequency spectrum: typical applications include guitar, strings, horns, cymbals or vocal-backing chords. In popular
music it can reinforce thin backing or enrich foreground. Other
interesting effects are found when it is applied with a 1±5 ms delay
to transients. It can also be used to resynchronize sounds that have
been produced a little out of step owing to the time a sound cue takes
to travel through air.
The short-delay effects in which phase is important were also originally explored in pop music by tape techniques before pure electronic devices were introduced. Two recordings were played,
initially in synchronization, and then one was slowed by gently
pressing a hand against the flange of the tape spool ± which is how
the resulting effect got the name of flanging. The recordings were
separated by a progressively increasing interval, causing phase cancellation to sweep through the audio range. This effect is now
achieved by variable digital delay, and is most effective when the
interval between the signals lies in the range from 50 ms to 5 ms.
Recombined sounds with delays in the 5±15 ms range have a different quality, but are still dominated by phase cancellation that is heard
most clearly when allowed to sweep through different frequencies. Phasing effects were originally explored by using analogue
[12.10] Flanging and phasing
The change in amplitude (vertical
scale) with frequency. A, The effect
of flanging ± here with a delay of
0.5 ms ± is evenly distributed
throughout the frequency range, but
is not related to the musical scale,
which is logarithmic. B, Cancellation
and enhancement produced by the
use of a phase-changing network:
the dips are distributed evenly over
the musical scale.
212
Reverberation and delay effects
phase-shifting networks, but these were distinctively different
from the `flange' effect in that cancellation was distributed throughout the audio range in an exponential manner (in the ratios
1:2:4:8:16, etc.) compared with flanging's arithmetical array
(1:2:3:4:5, etc.). The analogue phasing effect was therefore directly
related to the harmonic tone structure, unlike flanging's denser
overtone structure, much of which is not conventionally musical in
quality. With the introduction of techniques based on the use of
delay circuits, `phasing' reverted to the arithmetical pattern of
cancellation originally found in flanging, which many users still
preferred.
In a further variation on this, called chorusing, a signal in the
doubling range is also very slightly changed in pitch. In the midfrequency range this lends a feeling of depth to the sound; at high
frequencies it enriches the harmonic structure.
Other effects on offer might include differential delays for variously
defined bands or tones within some of the more complex DSP
programs, and move on from those to go even further beyond the
realm of natural sound modification.
Changing pitch
It is easy enough to change the pitch of a sound: simply make an
analogue recording (say, on tape) and replay it at a different speed.
But this immediately creates two problems. The first is that as pitch
increases, duration falls, and vice versa. The second is a radical
change in what the beginning and end of every component sounds
like. For example, if replay is speeded up to increase the pitch by
two octaves, duration comes down to a quarter and the characteristics of both attack and decay change so that the sound becomes
both far more percussive and drier than before.
[12.11] Tempophon
This was a mechanical precursor to
the digital pitch changer. It changed
duration and pitch independently.
The head assembly could rotate either
way:if the sense was opposite to
that of the tape, pitch increased; if the
heads chased the tape, it decreased.
Signal was fed from the head in
contact with tape, via a commutator
and brushes to a second recorder.
If this is acceptable, it can be used for simple effects. An early
example was a double speed piano composition. On a first recording
the piano was played with deliberation, to avoid the appearance of
inhuman virtuosity when speeded up. This was then played back at
double speed, and conventional accompaniment from the same
piano was added. The double speed piano sound was not just a
minor variation on ordinary piano quality: it sounded like a completely new instrument, and had to be treated as such in the musical
arrangement.
If effects like these are not the objective, it will be necessary to
change pitch and duration independently. Early attempts to do this
also used analogue recordings: the tape was read by a series of pickup heads on a rotating drum. Each moving head in turn sampled a
short length of the recording, and these were remixed. When the
pitch was raised, the result was composed of a series of segments
with tiny repetitions at every junction, or if lowered it would have
bits missing. This function may now be included as one of the many
Reverberation and delay effects
213
on a digital signal processor. For this effect it samples overlapping
or gapped segments from random access memory, and joins them
together. This process can never be perfectly natural: there are
always either bits of sound missing or bits repeated, however small,
so the quality achieved depends on the skill of the digital programmer. A good digital program can make the end-product much
smoother.
Whether it is done by the archaic analogue method or by a digital
pitch changer, the result also depends on the nature of the samples.
There is no problem with sustained unchanging tones: they just get
longer or shorter. But at the other extreme, transients may be clipped
or completely lost, or they may develop a stutter, all of which are
bad. There is a tendency for a gritty, grainy quality of sound to
appear. It is sometimes heard on commercials where the advertiser
wants to put in too many words (or to shorten a required product
warning), and then tries to cut its duration by more than it will
comfortably take.
Devices vary both in the smoothness with which they blend sound
and also in their capacity to avoid the worst defects. Experiment will
show the limits within which any particular change is acceptable: it
may be 20% in duration, pitch or combination of the two.
Some problems can be reduced by pre-emptive action. For example,
if the plan is to record and then change the pitch of an actuality
sound ± such as a foghorn in a busy seaway ± any background
recorded with it that cannot be filtered out will be treated as well.
Because it is natural, the background may not be noticed when heard
with the ordinary sound, but treatment can make it vividly apparent.
Plainly, for such applications, close microphone techniques on the
original recordings (to give good separation) are desirable. The
foghorn might then, for example, be used to construct a melody
from a sequence of pitch-transposed components.
There are all sorts of possibilities for special effects. A hollowsounding alien voice might be obtained by adding a delayed, pitchchanged repeat, while adding subharmonics creates a monster (such
as Star Wars' Jabba the Hutt).
At this point it is necessary to add a few caveats. One is that when a
singer's voice is subjected to a change that affects its duration, the
result can sound odd. This is due to vibrato, the cyclic variation in
pitch that gives colour and warmth to a singer's sustained notes ± a
technique that is copied on orchestral instruments such as the violin.
Analysis of voices that are generally considered beautiful suggests
that pitch variations of as much as 8% may not be excessive
(a semitone is about 6%). But the really critical quality of vibrato is
the number of these wobbles per second. The acceptable range is
narrow: for an attractive sound the vibrato rate must be between 5
and 8 Hz, and is independent of the pitch of the note sung. Outside
of 5±8 Hz the effect resembles either wow or flutter: the vibrato range
can be taken as separating the two. The direct consequence of this is
that only small duration (or speed) changes are possible without
taking vibrato outside its normal range. At worst, the result is to
214
Reverberation and delay effects
make the vibrato both obvious and ugly. At best, halving duration (and
doubling speed) produces the effect of a small, vibrant-voiced animal.
Tremolo, a similar variation on the volume of a note, is also dramatically affected by such changes; so too is portamento, an initial slide
into a note or from one note to another, which if slowed down
changes its character, or if speeded up almost disappears.
Another unwanted side-effect might be on the quality of male and
female voices. While these may differ in fundamental by an octave,
the formant ranges (i.e. the parts of the vocal sound that are emphasized by throat, mouth and nasal cavities and used to give the main
characteristics of speech) are, on average, only a semitone or so
apart, and vibrato ranges are also little separated. Small changes of
pitch therefore tend not to change the apparent sex of a voice, but
rather to de-sex it.
A more benign use of pitch change is offered by a program that
slightly spreads the nominal pitch as an anti-feedback measure. If the
treated signal is returned through a studio by acoustic feedback or
public address, the loudspeakers can be louder. Unlike the frequencyshifted feedback sometimes used for public address, this can be fed to
musicians, too, as the spread is centred on the original frequency.
Some of the (potential) problems associated with pitch change
(while leaving duration unchanged) can be sidestepped by proceeding directly to a pitch-correction program. This should be able to
take a sample of sound and shift it up or down as required, matching
it precisely to a chosen note on any of a wide variety of western,
ethnic, historical or invented-for-the-occasion scales. It will also
preserve intonation by defining a tonal centre, retuning that and
preserving the intonations around it. Variations on this allow the
creation of harmonies from an original (for example, to accompany a
lead vocal or `thicken' its quality) or to add vibrato.
The most sophisticated professional pitch and duration changers are
now extremely flexible, offering independently changing tempo and
pitch that can be mapped over an extended waveform or several
waveforms in parallel ± without loss of timing accuracy, stereo phase
or surround sound data (all possible dangers in digital processing).
This kind of process bridges a link between the range of DSP
functions that apply to all audio control operations as described in
this chapter and the much broader range that will be outlined in
Chapter 16, which introduces the virtual studio along with `plug-ins'
such as those offered on the Pro Tools platform and the MIDI
operations used in conjunction with them.
Sampling
Sampling and looping are further possible products of digital signal
processing, in this case bridging between delay effects and digital
audio editing (which is described in Chapter 19).
Reverberation and delay effects
215
Originally, sampling was achieved by physically cutting audiotape
and putting spacers on either side of it; to loop the sample, its two
ends were joined so that it would replay repeatedly for as long as
required. It could be played backwards or at a different speed. It
could be copied, introducing changes in volume or EQ. All sampling
and looping effects mimic this. Some MiniDisc and CD players offer
means of marking in and out points so that short samples can be
selected and replayed as loops, originally in order to help find an
exact point in the replay.
Digital developments from this are more versatile. Essentially they
at least start by doing the same things, but use digital recording,
editing and signal storage. Keyboards can be used for replay (and to
change pitch): they are widely used for percussion. And drum
machines use high-quality samples to build up repetitive sequences
that can be used either as a minimally-skilled foundation for otherwise live popular music or as the rhythmic underpinnings of a
synthesized composition.
Chapter 13
Recorded and remote
sources
All studios have a variety of additional sources that are mixed in
with the output from the studio microphones. These include disc
players ± for CDs, MiniDiscs and vinyl ± analogue and digital tape
decks (plus, for television, videotape and film sound), hard disk
sources including Zip drive and the output from digital editing
systems, and lines from other studios and remote locations, including ISDN and regular phone lines. Some of the same sources are
used in film and video dubbing theatres ± in fact, in any kind of
recording studio ± but their use is central to much of the output in
both radio and television production. This chapter deals with the
operational techniques for using them.
These are all programme sources comparable to studio microphones,
but differ from them in that they usually reach the console at or near
the much higher `line' level (zero level). Volumes may need adjustment, but no great preamplification. Most go straight to stereo
channels.
In radio, and to a lesser extent in television, each small segment of
this type of programme is collected, selected or pre-recorded individually. The creative process is spread out in time, and within the
limits of human fallibility and commercial pressures, each element
may have lavished upon it all the care it requires for perfection.
There is often a stage when most pre-recordings are complete or
require only last-minute editing, and all this `insert' material may be
taken to a studio to be broadcast with live links or re-recorded into a
complete programme. Another possibility is that it is scheduled into
an automated play-out.
In order to match levels, mix in effects or music, introduce fades and
avoid abrupt changes of background atmosphere, recordings are
mixed in to the programme, not just cut in `square'. Ideally, each
Recorded and remote sources
217
insert must be controlled afresh, so it is fully integrated into its new
setting, with special care for the relative levels at the beginning and
end of each insert (see Chapter 14). In the operation of replay
equipment, the most vital single element is good timing, because
where successive sounds are linked by mixing it is more difficult for
errors of timing to be corrected by editing. On live radio there is no
calling back miscues.
Digital automated broadcasting uses time code to control the output
from playlists and `jukeboxes' (which may just be different aspects
of the same play-out systems). These are capable of cueing the
routine transmission of many items in a random order, and can even
have overlaps programmed in. But manual techniques are still used
for much creative production work ± in fact, to build many of the
programmes or items that will later be broadcast automatically ± and
they also employ the techniques that, in their most developed form,
automatic systems must simulate. The fully digital alternative is to
assemble the material on hard disk using a digital editing system:
this can achieve much the same results but by more analytical
techniques. If a studio is still used at the end of this, it will likely
be for fine-tuning and review. Whatever the method employed, the
objective is the same, and is described here in terms of the studio
techniques. For audio editing, whether as part of the studio operations or carried out elsewhere, see Chapter 18.
Live-and-insert programmes
[13.1] Tape transport systems
1, Open reel (reel-to-reel). 2, Cassette.
3, Cartridge, a continuous loop.
Pressing the cartridge record button
automatically puts a tone (below the
audible frequency range) on the tape.
On replay, when this point on the
tape returns to the head, the tape
stops and is re-cued automatically,
ready for immediate replay. Tape cue
tracks may also have time code
recorded on them.
The linked actuality programme is typical of what might be termed
`live-and-insert' work: location recordings of effects and interviews
arranged in a sequence that helps tell a story. Unless each collected
item cues directly to the next, linking narration will almost certainly
be needed. Very similar in construction is the illustrated report, in
which summary, comment and studio discussion are mixed with
extracts from recordings of some event. Disc jockey shows, whether
of popular or classical music, may also be regarded as live-and-insert
shows. Here we can be sure that exquisite care has been exercised
in recording the `inserts': their arrangement within a programme
deserves care too.
In the following example, where the raw material is assumed to be
on tape, the script of a feature that combines recordings with live
narration might look like this:
1
TAPE
(BELLS. Recorded actuality: peal from beginning, 15'' and fade
under.)
2
NARRATOR
Bells. Bells to proclaim to the world that a new cathedral stands
ready for the praise of God.
3
TAPE
(HYMN, Recorded actuality: lose bells behind. Fade singing on
end of first verse, and out before start of second.)
4
NARRATOR
But this is a story which began one quiet night (Begin to fade in
FX) in 1940.
218
Recorded and remote sources
5
FX
(NIGHTINGALE. Fade in drone of BOMBERS until nightingale
is drowned out. Add distant ANTI-AIRCRAFT FIRE.
Hold bombers behind speech.)
6
NARRATOR
The bombers swept across the quiet countryside to the city. In the
city centre, the old cathedral . . .
7
FX
(BOMB screams down and explodes.)
8
NARRATOR
. . . a glory of Gothic architecture. Windows of richly stained glass,
set in delicate traceries . . .
9
FX
(BOMB explodes, tinkle of BROKEN GLASS.)
10
NARRATOR
. . . carvings and statuary that were the life-work of countless
medieval craftsmen . . .
11
TAPE
(Pre-mixed FX sequence: 2 BOMBS, begin to fade in FIRE and
hold, gradually fade out bombers.)
12
NARRATOR
. . . the hammer beam roof a triumph of human ingenuity . . .
13
TAPE
(FX continue: Several FALLING TIMBERS then CRASH as roof
falls. FIRE up to full, then fade to silence.)
14
NARRATOR
With the cathedral had gone many other buildings. The centre of
the city was largely a wasteland. Rebuilding could not be started
until after the war. But nothing could stop people dreaming of the
future . . . and planning for it.
15
TAPE
(Edited Interview: quick fade-in on atmosphere.) At first there
was quite a lot of opposition to the idea of a new cathedral: well,
of course, people said houses first and the new city centre, shops;
the cathedral could come later . . .
This requires two tape replay machines, a range of effects sources
and a microphone. During the preparation (`prep', or rehearsal)
stage, many of the cues on the console and play-in operators'
scripts will be overlaid with all the notes and numbers required to
complete it.
In fact, this example is at (or perhaps even a little beyond) the limit
of what is now put together without editing. An alternative and
perhaps more likely approach would be to supply the inserts as
components of a digital edit, to which studio narration is added by
recording it directly on another track. The editing is completed at the
back of the studio while preparation of other items continues.
Preparing recorded inserts
Each insert must first be considered as though it were a programme
in its own right: it must go through all the processes of rough and
fine editing that are needed to give it point and clarity. When carried
out `as live' this can often be as drastic as is possible when digitally
editing a programme. Anything that does not work well can be cut
out (but is still available), and the essence of confused or muddled
actuality can generally be interlinked with script to make it both
lucid and concise.
Recorded and remote sources
219
Heavy actuality background effects should be recorded for long
enough before and after foreground speech to allow for fades
(although if the effects, as recorded, are too short or sound wrong,
a loop or effects recording can often be used to overlap and provide
material to fade). Try to allow a few seconds of atmosphere for fades
even where the background is not heavy. An intake of breath before
a speaker's first word may be left on the recording. At normal
listening levels, neither atmosphere nor breath stands out, but they
do sound natural. Take particular care with any background atmosphere on speech inserted between music recordings on which studio
atmosphere is low.
If manual replay is envisaged, pre-recorded segments are assembled
to replay them in sequence, ensuring that any to be mixed or crossfaded are replayed from separate sources. At this stage a script may
give narration in full but inserts in cue form, with their durations,
like this:
NARRATOR
. . . spoke on his return to London Airport.
INSERT
Item 10 Dur: 27
In: What we all want to know . . .
Out: . . . or so they tell me (laughter)
NARRATOR
The minister stressed the need for unity between the two countries and
then said:
INSERT
Item 11 Dur: 70
In: It's all very well to be critical . . .
Out: . . . there's no immediate danger of that.
In this example there are two types of cue from narration. The first
needs about the length of a sentence pause, but the link ends with a
`suspended cue' that demands to be tied closely to the sentence that
follows. In the first case the atmosphere fade-up may follow the cue;
in the other it should go under the last words of the cue. In the
second case the timing also depends on the pace of the recording that
follows. For instance, `It's (pause) all very well . . . ' sounds odd if
tied too closely to the cue ` . . . and then said:'. Mark the peculiarities
of each cue in some bold private shorthand on the script. Again,
these suggestions are equally valid for play-in or for the digitally
edited alternative.
Other more complex overlaps between narration and insert might be
needed for artistic effect: how such fades and mixes can best be
arranged is described in more detail in Chapter 14. The object of
`prep' (or rehearsal) is to try these out, and unless the programme is
assembled by trial-and-error, or in short sections, cues may come up
so fast on the take that some quick visual reminder is needed for
each of them in all but the simplest sequences, and will be used as
a musician uses a score. There is no virtue in trying to play it from
memory.
The main things to mark on the script are:
220
Recorded and remote sources
.
.
.
.
.
the format and location of each source: which play-in source it
comes from (although on the console it is helpful to group inputs
from each format, so that a late change in which of a pair of
machines is used is not missed);
any oddity, peculiarities of timing or hazards to listen out for;
volume control settings and type of fade-in;
duration (or other notes on when the `out' cue is coming up);
type of fade-out.
Where the replay system allows precise, manual cueing, rehearse this
to get the feel for the timing. After cueing a replay there must often be
a tiny pause before fading up, to allow a mechanical device to run up
to speed: on analogue tape for example, this might otherwise be heard
as a wow, if only on atmosphere. Do not rely entirely on remembered
timings: different devices may have different run-up times, although
digitally controlled (time-coded) replay can be very fast.
Speech recordings with different background effects present a typical case for a mix. If the changeover point is internal to both
recordings, there are several ways of performing the mix, depending
on the length of pauses between words. In the simplest case, with a
full sentence pause on both recordings, the pause can be retained at
its natural length. Analogue tape is a medium that allows visual
control of the process ± by putting marks on both tapes to show
where any cross-fade should begin and end. Even if one of the
originals has no pause, it should still be possible to arrange a tidy
mix: for example, if there is no pause at the `out' point of the first
recording, then the second can be cued and faded up early. Where
there is no pause on either side, matching atmosphere can be filled in
from somewhere else on the recording.
In all cases match the levels with care: use the fader to `edge' one
side of the join in or out a few decibels to avoid a doorstep effect on
the voice. The same technique can be used where heavy atmosphere
ends too close to a last or first word.
Even if digital editing is used in the studio, for live transmissions some
operators still prefer to transfer the output to analogue tape before
going on air. Computers crash more often than tape replay fails ± and
the edited version still exists as backup. As a further justification, playout to tape offers a final chance to check the edit, a potentially vital
safeguard. And last-minute cuts can be made quickly on tape. However, the progress of digital play-out seems inexorable ± and is acceptable provided it is rigorously separated from other uses of the same
platform, and provides backup both of material and processors.
[13.2] Programmable CD player
For broadcasting studio use, a CD
player has a wide variety of functions,
including precision set-up at any
point ± with memory for a sequence
of many `start points' on different
discs ± and also looping and variable
speed.
Cueing play-in
A general-purpose radio studio will have a wide variety of play-in
sources, including perhaps two digital editing systems, three openreel tape decks (for backup or for on-air play-in), two DAT record/
Recorded and remote sources
221
replay machines, two cassette players, three programmable CD
players, a MiniDisc cart, two `gram' (gramophone or phonograph)
players, and perhaps a Zip drive record and replay cart.
[13.3] Care of compact discs
CDs should be handled with as much
care as vinyl discs. Treat them like
precision optics (which they are).
To avoid fingerprinting the tracks,
hold only by the edges or by rim and
centre. Remove any dust with a soft
air-brush. Avoid cleaning, but if a soft
cloth (or lens cloth) must be used,
brush very gently from centre to
edge, and not along the line of the
tracks. The only solvent that may
be used is dilute isopropyl alcohol
(and even this can lift the label).
For analogue open-reel tape, run the recorded sound backward and
forward over the replay head to identify the `in' point, mark that at
the head with a wax pencil, and then if there is atmosphere to fade
up, also mark the place from which the fade up will start. Ideally,
insert a 4-second yellow leader to end at this point and set this back
from the replay head far enough for the tape machine to run up to a
steady speed. Some professional machines do this quickly, in about
1 cm (1¤2 in), but in any case, if time permits, rehearse to check this.
On cue, replay and fade up from the first mark (or the end of the
spacer). The play-in machine fades up to full, with the console preset for the required (rehearsed) play-in level. Do not fade it in at the
console unless replay is cued by a remote start from there.
For DAT machines with `quick start' or `memory start', for nearinstant replay from the beginning of a recording, set up at the end of
the start ID: programme should follow in half a second. Any subsequent items on the same tape should also have their own start ID
(after a clear gap on the tape). As usual, check for any fade-up on
atmosphere.
A tape cassette also offers rapid replay: to set up at the end of a
leader, turn the spools by fingertip or a thick pencil to move the tape.
Commercial Compact discs (CDs) are designed to be run on domestic
machines only from the points programmed on to them, as listed in the
notes in the box. When setting up a CD on a standard replay machine,
allow time for these cues to be read automatically into memory, then
the track can be selected and held on `pause'. The delay after pressing
`play' is checked, and the track reset on `pause' until required. On
cue, it is started early enough to allow for any delay.
Studio CD replay machines can also be set to replay CDs and CD-Rs
from the start of a track in the same way. If `auto-cue' is selected, the
disc will set up for instant replay from the start of sound on the track
(but beware very quiet starts, as these have sometimes been clipped
by auto-start devices).
[13.4] Rapid-access CD rack
A home-made record rack for CDs.
This has a separate numbered spigot
for each disc. Each spigot consists of
two slices of wooden rod with a
screw through the centre to fix them
to a board that leans back a little.
The outer rod is about 14 mm (5 in) in
diameter, so that the hole in a CD
slides smoothly over it and rests with
its centre against the larger disc. The
spigots are about 14±15 cm (5±6 in)
apart. The side of the CD with printing
on is at the front and the recorded
side, touching nothing, is at the back.
It is easy to lift the disc by its edges
into the player.
For more flexible CD operations, begin by choosing the required
track, then press the appropriate buttons to select start and stop times
in the display window, followed by `auto pause' to activate the stop
time. If no stop time is selected, it should play to the end of the track.
There should be instant replay buttons to play several cues from the
same disc ± and the machine should also store a large number of
cues from different discs, and recall them when any disc is put in the
machine.
For precise location of a start point on the CD, it is useful to have
a search wheel. Play up to the required cue, and set the machine to
`edit': the search wheel then allows the sound to be run backward
and forward just like running tape over a replay head. When
a suitable point is found, press `edit' again to store it (or enter the
time, as before). The search wheel may also be operated with a
222
Recorded and remote sources
`loop' that repeatedly replays a tiny fragment, as it is shifted back
and forth until the exact point is found. `Loop' can also be used as
a sampler. Other functions of this CD player include variable speed.
[13.5] MiniDisc cart
An MD player designed to act as a
cart machine ± to play spot
announcements such as station ID,
jingles, trailers, commercials and
sports, news or other (usually fairly
short) recordings.
Before playing a MiniDisc (MD) cart, check any pre-sets, e.g. to
`single' replay. Insert the disc with its label up (an arrow on it points
`forward'). The name of the disc (or `no name') will appear in the
display window for a moment, then it will show that replay is set on
the first track; turning a knob switches this to the starts of other
tracks. Press `play' to check the start of the required track or `end/
monitor' to check the last 5 seconds; at any stage `standby/cue'
resets it to the same start ready to go again, and after playing the
whole track it will again reset to the selected start.
To start from a point within the MD track, play to somewhere near it,
then use the `search' keys: these repeat a short loop to find the
precise `in' point. Again, `play/pause' (from the chosen point) can
be used to check it, and `standby/cue' resets to it. The display
window shows elapsed time, or can be switched to give time to
end of track.
A MiniDisc cart machine can be used for spot announcements such
as station identification, jingles and commercials, as well as trailers.
A Zip drive cart, a single replay slot used for the same purpose, may
have 10 cues ready for instant replay, each on a numbered button,
with others (in double figures) almost as easy to find and run.
A tape cartridge (if one should still be encountered) is made ready
to start from the beginning by simply pushing it into the slot.
Cartridges contain loops of lubricated tape in standard lengths of
say 10, 20 or 30 seconds, to allow various durations to be recorded.
They switch off if there is a pause of more than, say, 2 seconds,
after which the tape carries on until it has returned to the starting
point, to await re-cue. Cartridges were once widely used in multislot carts where their automatic reset was useful in run-and-forget
operations.
[13.6] Simple arc scale locator
This can be used to help find a
particular groove on a vinyl or archive
disc. The scale rests over the
turntable spindle at one end and in
the hand or on a block at the other.
It can be made very simply from card.
Some record players have optical
systems to display a scale in a nearby
window.
`Grams' (phonograph players) are used for vinyl or archive discs.
A fairly brutal way of setting up is to run the disc a little beyond the
`in' point, stop and run back manually with the needle in the groove:
this was the method actually used in the years before it was adapted
(less destructively) to running tape back and forth over a replay
head. Some turntables have groove locators that show the position of
the arm on a scale in an adjacent window. A strip of plastic or card
with a scale extending across the grooves can also be used ± though
both methods can only give a rough indication. On vinyl discs,
successive excerpts can be found reasonably quickly by lifting and
relocating the head. It helps that on a grooved disc, variations in
volume are clearly visible.
If several discs are to be played, they can be numbered with a wax
pencil on the smooth part just outside the label. In the days of 78s it
was even possible to mark the actual groove with a wax pencil
sharpened to a knife-edge. If it is driven from the centre, the turntable acts as a flywheel, to even out speed variations: `flanging'
Recorded and remote sources
223
effects were originally created manually by resting a finger against the
edge of one of a pair of identical discs that had been running in sync.
During replay, there is always a danger that a pick-up will run
forwards (a groove run) or, worse, back. If a repeating groove occurs
during a radio transmission, one cure is as violent as the effect: fade
out, jump the stylus on a little, and fade back up. Alternatively, as the
stylus may require little persuasion to return to its original path, or at
worst to jump forward, simply rest a pencil very gently across the
pick-up arm and `squeeze' it towards the centre of the record.
Archive disc recordings in outdated formats are transferred to a more
suitable medium before use. Use vintage replay equipment or a
modern replica to convert the recording into an electrical signal
without unduly wearing the original, and follow this by any treatment that helps to clean up the sound: there are DSP programs for
this (but prized `vinyl quality' may also be diminished by digital
treatment).
Returning to more modern techniques, hard disk replay employs a
recording stored as a digital file in computer memory ± often in a
digital editing system. This allows instant replay, and can be set to
run a series of overlapping files. For digital editing, see Chapter 18.
Digital audio files can also be cued in arbitrary order from a MIDI
keyboard controller. In principle, MIDI devices can readily be
combined with other digital operations such as sampling (re-recording between pre-set cues) and pitch changing. They could be used,
for example, to create a tune out of a single toot on a car-horn, or
to comply with a director's request to `make the footsteps a bit
deeper'. They are therefore particularly useful for sound effects
(see Chapter 15).
Remote sources
Another type of source that may appear on a console at a radio or
television station is a line from some other studio or remote location
(in Britain these are called outside sources). A local radio station
might have lines to the regional weather bureau and recreational
areas, local government offices, traffic monitoring points and vehicles, sports arenas and concert halls, together with other information
centres for news, business or entertainment reports. International
links, network lines, other local or distant studios and the telephone
system may all be used. Contact between the central studio and the
remote source may take various forms. Often there are two lines: the
programme circuit, which may be one- or two-way; plus a control
line, which is effectively a telephone link between control points at
the two ends.
For many links of this kind, the programme material is on a broadband digital telephone line (ISDN). Both remote source and the
studio have matching codecs. When ISDN is used for broadcasting,
the greatest potential hazard is the quality of the headset and codec
224
Recorded and remote sources
at the far end, but in this application, the broadcaster can usually
ensure that they match, and any minor loss of quality due to data
compression is acceptable. The studio usually has a calling unit with
a keypad, display window and a few extra buttons. Calling is similar
to off-the-hook phoning: key in the number, then press `dial'. In the
window, `ISDN line' appears, followed by confirmation (or not) that
the codecs match, which means that the line is working both ways. It
is then ready for diversion to the console as an outside source. In case
error message numbers appear, it is best to know in advance what they
mean: mostly they refer to the usual telephone dialling problems.
Note that a codec with greater bandwidth (but still with `acceptable'
data compression) is sometimes used for music. If the studio does
not have the necessary unit, it might be available in a switching
control area outside the studio: the number can be dialled from there
and the line fed to the studio, again as an outside source.
Generally, any kind of link is set up and checked before it is diverted
to the console. In the case of a one-way music line (i.e. broadcastquality line) plus control circuit, in order to prove the line, contact is
established first on the control circuit, and the distant contributor is
asked to identify the broadcast line by speaking over it and also to
send line-up tone. If the programme is not yet on air, these can both
be checked with the channel fader open, so that the level of the
incoming material can be accurately set and its quality established.
If the line is not available until the programme of which it is to form
part is already on air, the line can be fully tested to the last switching
point prior to the active part of the control console. On its assigned
channel within the mixer it is checked on pre-hear (also called prefade or PFL) using a small loudspeaker or headphones.
A control line may be double-purpose: as well as being used for
discussion between the two ends before, during and after the contribution, the return circuit may have cue programme fed on it. Also,
the producer's talkback circuit might be superimposed on the cue
programme. The cue feed is checked by the contributor as part of the
line-up procedure.
Sometimes the outside source is a `self-drive' studio. If there is a
separate, manned control point the cue programme may terminate
there and the contributor may be given a cue to start by the operator.
Alternatively, it may be fed to the studio for the contributor to hear on
headphones or (for television) a concealed earpiece ± better than a
loudspeaker, which might cause coloration. The feed might be for the
beginning and end of the contribution or, in an interview, will
continue throughout. If there is no alternative to a loudspeaker being
on while the microphone is open, the contributor could be asked to use
the directional properties of both to minimize spill, and then reduce the
loudspeaker level until it is just clearly audible at that end. When
the contributor speaks, it is possible to test the circuit by raising the
incoming level until coloration is heard: this establishes a level to keep
well below. During the link, try reducing the volume of cue feed while
the contributor is speaking, but take care to give no reason to fear that
it has been cut off. Anti-feedback pitch spreading can also help.
Recorded and remote sources
225
Another arrangement is to have broadcast-quality or broadband lines
both ways. This is necessary if the material is to be broadcast or
recorded at the same time at both ends.
In practice it may be unwise to go directly on air with members of
the public whose language or manner are unpredictable. A digital
editing system can be used `on the fly' to mark and delete unacceptable
language and select a good out point for a `near-live' contribution.
Clean feed and multi-way
working
The simplest form of cue programme is the output of the mixer
studio. This is satisfactory if just a cue to start is needed. The remote
source then switches to hear local output until the contribution ends.
In the more general case, participants will hear the originating studio
and contributors from other sources throughout. Some speakers like
to hear their own voices coming back ± it may help them judge their
delivery and effectiveness ± but most do not; at worst it may
seriously inhibit their performance. Ideally, all speakers who hear
cue on headphones should have the choice of hearing themselves
or not.
The choice is not always possible: if, say, speakers at the New York
end of a transatlantic discussion were fed the full programme from a
mixer in London they would hear their own voices coming back,
delayed by the time it takes for the signal to travel there and back.
By cable, at the speed of light, this is a thirtieth of a second, bad
enough to make speech decidedly uncomfortable, and if the return
signal is loud enough, it becomes impossible to carry on without
stuttering. Digital conversion can also introduce delays that do this.
Satellite links are worse, with quarter-second delays in each direction. For television, links with delays may be used even for contributors in the same town.
[13.7] Clean feed circuit
For multi-studio discussion, speakers
in the studio can hear a distant source
on headphones without hearing their
own voices. `Clean' cue is also
provided. 1, Microphone and
2, Headphones (or concealed hearing
device for use on television) for a
speaker in the studio. 3, Clean feed to
distant studio. 4, Combined studio
output. 5, Incoming line. This
arrangement is essential for both
radio and television when studios are
linked by satellite, which introduces
a long delay to any sound that is
returned to its point of origin.
For these it is essential to use clean feed (also called mix-minus), in
which participants hear everything except their own contribution.
Most consoles provide circuits for this. For two-way working where
the outside source line is added to the studio output after the main
gain control, a clean feed can be taken off just prior to the
main control and fed via a parallel fader: in effect, there are then
two main controls, one with and the other without the remote contribution.
A typical use of clean feed is to provide an actuality feed of an event
to which various broadcasters add their own commentary. The
actuality sound is fully mixed at the originating console, where
commentary for the home audience is added through an independent
channel fed in to the parallel output. This is similar to the provision
of mixed music and effects tracks in film dubbing: there, too, a user
may add commentary separately, perhaps in a different language.
226
Recorded and remote sources
Another possible arrangement uses auxiliary circuits. Each channel
in the mixer has provision for a feed to be tapped before the fader
and fed to an auxiliary bus, where the output of all channels selected
is combined in a rough mix that can go to any local or remote point.
Many consoles have two or more auxiliary `sends', to allow multiway working.
[13.8] Multi-way working
In this example, four remote sources
each receive a broadcast-quality mix
without their own contribution. This
matrix can be achieved by the use by
`auxiliary sends' or `independent main
output' facilities in the central mixer
console.
An application of this is the `independent evidence' type of programme where a presenter questions two people in separate studios
who cannot hear each other (e.g. to examine the reactions of various
partners who know each other well: brother and sister, secretary and
boss, or husband and wife). Three studios are required: the host,
listening on headphones, hears the full programme output, while
in two remote studios the `victims' are fed nothing of the other
person's answers, and only such questioning as is directly addressed
to them.
A complex form of clean feed is used when recording sporting and
other events of international interest. Picture and sound are
recorded on videotape and a synchronized multitrack sound tape.
Commentaries in different languages can be recorded at the same
time and recombined with the actuality sound when the tapes are
replayed.
Telephone circuits in radio and
television programmes
Prior to the introduction of ISDN links, telephone circuits were not
routinely used in all broadcast programmes for several reasons:
.
.
.
.
Speech quality from the handset is good enough for communication but far below that normally required in broadcasting. Distortion and noise levels are high, with an uneven frequency
response.
The standard telephone bandwidth is narrow, 300±3000 Hz in the
USA and 300±3300 Hz in the UK. This is enough for full intelligibility, but, again, less than is desirable for broadcasting.
The telephone system cannot be expected to guarantee the freedom from interruption needed on lines used for broadcasting.
Connection by automatic equipment at a given time cannot be
guaranteed: junctions, overseas circuits or the number may be
busy.
Some of these conditions still apply where ISDN broadband is
available. Plainly, since broadcasting is geared to different technical
standards, telephone circuits are used only for good reason. Valid
reasons are:
.
Extreme topicality. There is no time to arrange higher quality
lines or to get a recording or the reporter to the studio, and the
same programme value could not be achieved in any other way.
Recorded and remote sources
.
.
[13.9] Broadcasting live telephone
conversations
1, Remote telephone. 2, Studio
telephone. 3, Constant-volume
amplifier. 4, Studio microphone.
5, Console. 6, Studio sound output.
.
227
Extreme remoteness. Here, physical inaccessibility is the problem
± a distant country from which better quality lines are not available. The speaker may be in an area that has been struck by
natural disaster. (But satellite links can still often be set up in
such circumstances.)
Audience participation. The ordinary citizen at home, without
elaborate preplanning, is invited to take a direct part in proceedings to which the audience is normally only a passive witness.
The phone-in. This is an extreme example of audience participation, in which telephone quality is an intrinsic part of the show.
In practice, after the circuit is established on an ordinary telephone ±
often one of several set up in the studio ± a feed is diverted through a
`telephone balance unit' to the studio console, where it is mixed with
studio output to add the full frequency range of the near voice.
The use of a constant-volume amplifier or (essentially the same
thing) a combination of amplifier and limiter will ensure that
incoming speech is as loud as that originating locally. Note that
the gain of a simple limiter should not be set too high ± say, to a
maximum of 8 dB ± or line noise may be boosted noticeably during
pauses in the conversation. However, if there is a noise gate in the
circuit (and assuming there is adequate separation between speech
and noise levels) the gain of the limiter can be substantially
increased. Also, when the call is checked for quality before transmission, it is essential to establish that no crosstalk is intelligible.
The call remains connected as long as it is routed through to the
studio, even if the studio telephone line is not open, and should not
be disengaged at the end of the conversation when the studio line is
closed. It is finally disconnected after switching back to the control
phone or by the remote caller replacing that receiver.
For the broadcaster, special services may be arranged: fixed time
calls may be booked, or timing pips that would normally be superimposed on a call may be suppressed. From the point of view of the
engineers of the telephone service, there are two rules that must be
observed:
.
.
no equipment should be connected that might feed a dangerous
high-voltage signal to the line;
no unusually loud signals should be transmitted as these could
inconvenience other users, inducing abnormally high crosstalk.
The way in which calls are used in programmes may be subject to
a code of practice agreed with the telephone company. Whether the
following are seen as rules or just as matters of common courtesy
depends on the local arrangements and on regulations imposed by
the licensing authority. These may vary from country to country and
also with the type of use (recording or live broadcast) and to a lesser
extent the kind of programme. Some are common sense; others are
designed to avoid complaints to co-operating organizations whose
service is inevitably seen at a technical disadvantage; yet others are
intended to limit intrusion.
228
Recorded and remote sources
The broadcaster should:
.
.
.
.
.
.
.
Obtain the consent of the other party before a broadcast is made.
Avoid using party lines.
In the context of a call, avoid broadcasting comments that might
sound like criticism of the service provider.
If calls become unintelligible or subject to interference, terminate
their broadcast as quickly as possible.
Immediately abandon calls that are interrupted by an operator or
crossed line. Someone should then call the other party as soon as
possible to explain what happened.
Broadcast telephone numbers, or permit them to be broadcast,
only by prior arrangement. Measures may be needed to avoid
exchange congestion and prevent emergency calls getting
blocked ± this may prove expensive.
Avoid involving operators or other telephone service staff in the
broadcast.
Where a conversation is recorded, the same result must be achieved
as if the item were live, except that editing may be used to achieve it.
In order to ensure that quality is as high as possible, speakers at both
ends may be asked to speak clearly but without shouting, not too
fast, and with the mouthpiece by the mouth, not under the chin.
Broadcast sound should not be loud near the telephone. If there are
parallel extensions at the caller's end, these should not be used by
others to listen to the call, as this too could reduce the quality.
Broadband digital links may be used if the speaker at the far end has
a suitable headset and codec. A public or company office, for
example, may have the equipment ± though it may turn out that
the quality is not as good as when both ends are controlled by the
broadcaster, in which case ordinary telephone quality may be better
than some peculiar compromise. If ISDN is used, it is set up as
described earlier.
Chapter 14
Fades and mixes
This chapter covers what may seem to be just a few tiny details of one
of the simplest operations that any console (or virtual studio) user can
face. But it is more than that. In fact, as much as anything else, it is the
way fades and mixes are carried out and timed that characterizes a
polished, professional product. Not every fade should be just a slow,
steady slide in or out. Everything from disc jockey shows to high
drama on film or television depends on a sensitive hand: a good fade
is like a well-played piece of music. A good operator does not wait for
a cue and then act on it, but rather, anticipates it so that cause and
effect appear to coincide ± which is, after all, the purpose of rehearsal.
What constitutes a smooth fade? The ear detects changes of volume on
a logarithmic (decibel) scale, so all faders are logarithmic over much of
their 30±40 dB working range. Below that, the rate increases ± again,
roughly in line with our expectations. At the lower end of a fade, as the
listener's interest is withdrawn from the sound, the ear accepts a more
rapid rounding off. The fade is, in fact, a device for establishing a
process of disengagement between listener and scene: once the process
is well established, it remains only to quickly wipe away the tail end.
An even movement on a standard fader does most of this, but the rate of
fade in or out can often be increased at the lower end, and reduced to
allow fine control in the final few decibels at the upper end.
A more complex use of faders overlaps their operation, as a mix.
(Music mix-down, a separate subject, is described in Chapter 9.)
The fade in radio drama
Consider first the fade in radio drama. For many users, this
must seem an almost forgotten application. If so, think of it as a
model for the more general use of operational techniques ± and
230
Fades and mixes
not just in all kinds of radio production, but throughout all uses of
sound in all media.
Radio drama often adopts a convention that each scene starts with a
fade in and ends with a fade out; narration, on the other hand, is cut off
almost square. This ensures that there is never any difficulty in
distinguishing between story and teller. Within the story, the type
and rate of each fade and the length of pause that follows it all carry
information for the listener. The example of radio (and particularly
how dramatic effects are achieved in radio) is useful because in
creating a sound picture the resources available are much more slender
than those for film or television pictures, which can employ cuts,
fades, dissolves and wipes, plus all the elements of camera technique.
A great deal has to be conveyed in radio by subtle variations in a fade.
For a scene change, the gentlest form is a fade to silence over 5 or 6
seconds, a pause of a second or two, and an equally slow fade-in.
This implies a complete discontinuity of time and action. Faster
fades imply shorter lapses of time, smaller changes of location, or
greater urgency in storytelling. For instance, for `Maybe it's down in
the cellar, let's go and see . . . well, here you are, have a look round',
the fades can be quickly out and back in, just enough to establish a
tiny time lapse.
Moving off, then back on microphone, perhaps helped by small
fades, is used for still smaller changes of location, say from one
room to the next. It is convenient to assume that a microphone is
static within the field of action that may be moving, e.g. a scene in a
car. Any departure from this convention must be signposted by
dialogue or effects, or the audience will be confused. A scene
starting by a waiting car that then moves off may create a moment
of confusion if it is not clear whether the microphone goes with the
car or not. Dialogue pointers are often used to clarify this, and their
deliberate misuse can raise a laugh:
[14.1] Fades in drama
Above: Simple fades between
scenes, with smooth fade out then
smooth fade in. Below: Fades
between narration and scene. Slight
edge out at end of narration (A) and
smooth fade in (B). In these diagrams,
the baseline represents silence and
the upper limit is normal volume.
FUNNY MAN
Quick, drive to Tunbridge Wells!
FX
(High-powered car departs and fades out.)
(Pause.)
FUNNY MAN
He might have waited for me to get in.
Fades and mixes
231
To some extent, moves on and off microphone are interchangeable
with fades. In one survey, listeners found that a fade gave as good an
impression of moving off as did actual movement. It only works that
way round: if a move is used to represent a fade, the shift in acoustic
perspective makes it sound odd ± unless it is deliberately exaggerated, perhaps by combining a fade with increasing AR to suggest a
move off in reverberant surroundings.
How deep should a fade be, before it is cut off? Plainly, a scene
that has a strong curtain line cannot be faded much, and a new scene
with a first line that contains important information needs a fast
fade-in. For both of these the adjacent fade must be conventionally
slow, or the intended scene change will sound more like a pause
in conversation.
Difficulties arise when a scene begins with what on stage would be
an actor walking in: in radio, this is reinterpreted using dialogue,
acoustics and effects to establish a scene, together with deliberate
fades for exits and entrances within it. This may mean breaking a
fade into several parts: it is no use fading on a pause. So, if an actor
has two separate lines while approaching the microphone, the move
in might stop for a moment during any intervening line. Similarly, it
is better to hold up a fade-in momentarily if there is a gap in speech.
It may even be necessary to reset, jumping up or down in level
between lines from different speakers to make the overall effect of
the fade smoother.
`Door opens' is another instruction that may head a scene. For such an
effect to be clear, it has to be at close to normal level, and the fader is
then taken back a little for the fade-in proper. In general, a short spot
effect of this kind is not recommended for the start of a scene.
Educational radio may employ shallower fades, partly because of the
listening conditions in classrooms: a poor receiver or boomy room
acoustic. Producers who visit schools to listen with their audiences
recommend high intelligibility, low dynamic range and background
effects held well down so they do not interfere with speech.
Listening conditions in some parts of the world are poor because
crowded short or medium wave transmissions are heard at the limit
of their range, or on small, cheap radios. For these, too, a form of
presentation that uses shallower fades is necessary.
[14.2] Complex fades
Above: A, Speech. B, Silence.
Do not fade in on silence. Below:
Complex fade to accommodate a
quiet effect at the start. At A, fader
is open for `Door Opens', then taken
back for fade-in (B) on speech.
232
Fades and mixes
Mixing in effects
If the convention is that each scene opens and closes with a fade,
plainly there must be sound to fade on. Using unimportant dialogue
for this is dramatically weak. To fade out on laughter (where appropriate) is better, but using background effects is of more general use.
As the scene comes to a close, speech is faded down and effects lifted
to swamp the line; then they too can be slowly faded out. Similarly, a
continuous background of sound can be used for a fade-in ± effects are
good for establishing new localities ± and here again the sound may
be peaked for a moment, then faded gently as dialogue is brought up.
For another kind of fade-in, bring up effects a little before speech ± just
enough for it to be clear what they represent, but not enough to draw
attention away from the first words. The more complicated the background effects, the more they will need to be established first. In either
case, they must nearly always be dropped to a lower level behind the
main body of the scene than at the beginning and end. The level then
may have to be unrealistically low, and lower in mono than in stereo,
where the spread background interferes less with intelligibility.
Varying the level of effects behind a long scene helps to stop
the sound getting dull: it should be lowest behind speech that is
important to the plot, and louder when dialogue refers to the backing
sound, or when characters have been directed to pitch their voices
against it. Performers should be encouraged to sharpen their tone if
in similar natural circumstances they would do so.
Effects can often be peaked as a bridge between two parts of the
same scene: different conversations at a party can merge into and
emerge from a swell of background chatter. In film terms, this is like
either a slow camera move from one group to another or like one
group moving away and others coming closer to a static camera.
When you construct a scene for radio it is all too easy to imagine that
you are being more precise in depicting movement than is actually
the case. Fortunately it is not usually important if listeners see it
differently from you. Nevertheless, it may be possible to be more
precise, perhaps by giving the impression of a tracking shot by
peaking effects and then gradually changing their content.
[14.3] Fades with effects
Above: Establishing effects (A)
before speech (B) on separate
faders. Below: Fading up
effects (A) with speech (B) and
then holding them back to avoid
an excess of distracting backing
sound.
Fades and mixes
233
[14.4] Mixing effects
Background effects (B) can often be
varied continuously and can
sometimes be peaked to bridge
scenes, here as the fader for speech
(A) is dipped.
Beware the overuse of background effects. Used intermittently, they
enliven a production; used for every scene, they become tedious, and
more so if they are too loud.
Fades in television and film
For television and film the rate of fade and timing of any pause
is normally set by picture: in the simplest case a fade of picture
is matched exactly by fade of sound, but when there is low but
significant sound during the fade that should be emphasized. Similarly,
an effect of little significance that sounds too loud can be reduced.
There are plenty of obvious uses for effects and music during picture
fades. For example, to bridge a time lapse in a party scene, peak the
chatter background rather than fade it out: this firmly implies that
action and location are unchanged. If the new scene then turns out to
be a different party, a deliberate incongruity is introduced. A change
of scene from one noisy location to another might be made by
peaking effects against the fade and mixing while the picture is
faded out. This works best if the two sounds are characteristically
different. Such techniques should be used only with calculated and
deliberate intent. It is self-conscious in that it draws attention to
technique, at least for as long as it takes to establish a convention.
The dramatic reason must be strong, with sound that is important to
the action.
The less obvious use of effects, as described for fades in radio, also
has its uses. When dissolving between pictures it is normal to mix
sound backgrounds (or fade the louder of them in or out) to match
picture. Here a peak in effects to bridge the mix may be accepted.
More interesting, because it is more open to the creative use of sound,
is the picture cut from one location to another. The simplest case
uses a straight cut with dialogue (or strong action) to carry it. For
example:
Vision:
MS oilman
. . . turns to see distant figure striding
towards burning rig
Paper in manager's hands, fast tilt up
to CS face.
Sound:
FX: Burning oil rig
OILMAN (V/O): Sure it's easy. When you've
stayed alive as long as this boy has, the hardest
job is working out what to charge the client.
MANAGER: Nine and a half million!!
234
Fades and mixes
Here the effects behind the first scene must fade out fast. The next
example uses a variety of techniques on three shots:
Vision:
Through arch to cathedral tower
. . . tilt down to boys approaching
under arch and past cam
Sound:
Distant choir sings
Chatter, footsteps
NARRATOR: In the shadow of the medieval
cathedral this group's interest in electronics
complements traditional activities like choral
singing . . .
WS playing field, cathedral and school
buildings b/g, pan L as ball and players
cross diagonally towards cam . . .
when centred on science block,
z/i past players to windows
Wide HS in lab, 2 boys and
glassware f/g
RUGGER PLAYER: Break!
NARRATOR: . . . and Rugby football.
Perhaps rather surprisingly, the work we're
going to see started not with physics, but here
in the chemistry lab.
BOY: Oh no! If we've got to do this, it's set all
wrong . . .
The picture has been shot with the commentary in mind. The music
was recorded separately; but chatter and footsteps, football effects
and laboratory sound were all recorded in sync with the picture.
Before mixing, the chatter and footsteps track is displaced a few
frames because the approach then matches better with the tilt down.
The music is held low from the start and chatter and effects fade in
with the tilt down to dominate it. The shout `break' cues the cut
to the football field and kills the music. The football effects that
follow on the same track are therefore brought up to full volume
before the picture cut. With the first zoom, the effects dip sharply
to a level matching that of the laboratory, so that a quick mix on
the cut is enough.
The key device is that of sound anticipating picture. Here the result
seems natural, rather like hearing a noise in a street and turning to see
where and what it is. This can also be used for strongly dramatic cuts.
More noticeably to the viewer, the anticipation of sound can be used
stylistically, as when a football fan who is talking about some big
match lapses into a daydream during which we hear the sounds of the
crowd distinctly as though in his thoughts, followed by a cut to the
crowd scene at the match and the roar of the crowd at full volume.
Dialogue overlapping picture is now a widely used convention, but
needs care as it can call attention to technique or, worse, cause
confusion.
A whip pan generally implies a spatial relationship, which may offer
interesting possibilities for the use of sound, as here:
Vision:
LA through rigging to flag on mast
Sound:
LA figurehead of Nelson,
band crossing R±L
NARRATOR: . . . and the martial music, the
uniforms, the discipline and the impressive
reminders of a glorious past,
Music: military band
starts playing
Fades and mixes
MS Nelson figurehead
. . . whip pan R and down
Cut in to matching whip pan to boys
turning capstan
Closer, pairs of boys push spokes of
capstan through shot
MS boy, boat in b/g hauled up
slipway
235
all tell us that this school trains its boys for
the sea . . .
FX: capstan turns,
ropes creak, marching
Their project, too, is suitably nautical.
BOY: With two of our boats at school we
had a problem. . . .
In this case the whip pan is not a single shot of the two objects: the
figurehead and the capstan are half a mile apart. That they are all
part of the same general location is suggested by a series of atmospheric detail shots with sound overlaps. The sync music from the
second shot has been laid back over the first (mute) picture and also
continued through the whip pan, but drops sharply during the camera
movement. On the capstan shot, the associated effect (recorded
separately) is allowed to dominate, and the music is eventually faded
out under the boy's speech. The noise, relevant to his words and to
subsequent shots, is held low behind his voice.
The use of sound in all these ways is suitable to both television and
film. So, too, is the use of music links between scenes. Indeed, the
music bridge is a device common to radio, television and film.
Theme and background music
Few things about broadcast drama and documentaries attract so
much controversy as the use of music. This is because, to be
effective, and perhaps also to justify having specially composed
music, it is often loud. In contrast, in the home, many people like
speech to be clear, with less and quieter music (cinema audiences
seem less likely to complain). Most people agree that, when it
succeeds, music helps create works of high quality; but at worst it
may just be used to smudge a supposed elegance over a scene that
would be as effective with natural sound.
When television shows works that are the product of, or derived in
style from, the older traditions of film, music is still `in'. But to meet
the taste of home audiences, simpler styles have been developed,
artistically justified and found to be effective. This is linked to a
greater use of well-recorded effects: these either increase realism
or enhance emotional response, just as music does. It follows, more
than ever, that music must be selected with great care (or rejected
if there is not a perfect fit) and integrated into programmes with
delicacy.
The ideal is either specially composed music or well-chosen selections from commercially released records. Sometimes suitable background music can be found in `mood' record catalogues. But use this
with care: it is easier to lose tension and coarsen texture by adding
music than it is to enhance it. When in doubt, avoid it.
236
Fades and mixes
[14.5] Pre-fading music
The pre-fade time is the number of
seconds from the scheduled end of
programme (in this example, 77
seconds). A, Last item of programme.
B, Closing music is playing but not
faded up. C, Announcer. D, Music is
faded up to close programme.
Signature tunes and theme music must be in harmony with, or provide
a counterpoint to, the qualities of the programmes they adorn. At the
opening they must suggest the mood of what is to follow, and at the
end, sum up what has gone before. Another quality often sought in
linking music is `fadeability', with points at which it can be got in and
out tidily, all this in a piece of the right duration. At the closing, to
reflect the progress of the story, the selection might be a more
complex development of the music at the start. For some pieces, this
means (conveniently) that the end of the record can be used to end the
programme. But it is not so common for the start of a record to
provide a good crisp opening signature; more often it is necessary to
disregard any introduction and start clean on a main theme.
For the end of live radio shows, closing music is often pre-faded: this
is a device used to ensure that a slot is filled exactly to the last second.
The term `pre-fade' here means `not yet faded up'. (An American
term for this is dead-roll.) In a typical radio pre-fade it was known
that the duration from an easily recognizable point to the end of the
record was 77 seconds. So the replay was started that same 77 seconds
from the scheduled out point. Usually no more than 10±20 seconds
was actually needed, and this last remaining part of the record was
faded up under the closing words, or a little earlier if a suitable phrase
presented itself to the operator, who listened on headphones.
Mixing from speech to music
Some of the ways in which music and speech may be linked together
in a record show are outlined below. In this example, linking words
cue a pop vocal that has a short instrumental introduction:
.
.
.
Straight ± or nearly so. The `intro' on the record is treated as a
music link between introductory words and vocal, of which the level
can often be altered to match the speech better. After this link, the
record may be faded to a suitable overall level. The first notes of an
intro can sometimes be emphasized to give bite to an unusual start.
Cutting the musical introduction. The first word of the vocal
follows the cue quickly, as though it were a second voice in a
brisk argument. It may be necessary to edge the record in a little.
Introduction under speech. The start of the music is timed and
placed at an appropriate point under the specially written cue.
The intro is played at a low level, so it does not reduce the
intelligibility of the words, but there may be a gap in the cue in
which the music is heard on its own for a few seconds: it is
Fades and mixes
237
[14.6] Mixing from speech to music
A, Speech. B, Pause in speech.
C, Music `intro'. D, Music at full
volume. During the pause the music
is started and then dipped behind
continuing announcement. It is finally
lifted on cue to its full volume.
established. The fade-up starts just before the final words, to be at
full volume for the start of the vocal, which may follow the cue
after a breath pause.
A coarser technique employs compressors arranged so that speech
pushes the music level down: an American term for this is ducking.
With good-quality reception, the effect is unattractive, but it is suited
to domestic radio background listening or car radio reception in
noisy conditions, and it is operationally simple.
When links are recorded in advance to a play-out system the same
type of links can often be programmed in. The final 10 s of
the outgoing record is played to headphones for cue purposes, the
linking words are recorded with a programmed cross-fade and the
process is repeated on the link to the next record. The play-out
system stores the links separately and reconstitutes it all, including
the fades on transmission.
Joining music to music
The simplest music link is the segue (pronounced `segway' and
meaning follow-on), in which the new number follows either after
a very short pause or in time with the previous piece. Sometimes an
announcement may be spoken over the intro at a convenient point, or
over the join.
In some cases it may be better to avoid endings and introductions by
mixing from one piece to another. Each 2- or 3-minute item on a pop
record is, naturally, written and recorded as a complete work in
itself, with a beginning, middle and end. In a fast-moving record
show only the middle may be wanted, or the intro (or ending) might
be too big for use in the show without throwing it out of shape. If the
rhythm is strong in both pieces, precise timing is essential. It is
easier if the music reaches a point where its form is not too definite;
where, for example, the outgoing passage can be faded on a sustained
note and the next brought in on a rising arpeggio or glissando. But
while all this can be fun to put together, it may not give the audience
much added listening pleasure: do not mess records around just for
the sake of it.
A segue or mix is possible only if keys are the same or suitably related.
It may be that keys that are nominally the same are not close enough in
pitch without some adjustment. Expert musical knowledge is not
needed, but good relative pitch is essential. Many replay machines
have provision for fine adjustment of speed, and so pitch.
238
Fades and mixes
In any record programme, too, lend an ear from time to time to the
key sequence: a lively variety of keys adds vitality. The question is,
how to get to a remote, unrelated key. Within a piece of music it is
generally done through a progression of related keys. That may not
be possible when linking pre-recorded items, so unless a dissonance
is actually wanted, some other means of transition must be found.
There is a story of a music-hall conductor who, when faced with the
problem of a quick modulation, simply wrote in a cymbal crash and
promptly jumped to the new key. The sudden loud noise of indefinite
pitch was designed to make the audience forget the key that went
before. In radio a marginally more subtle method is used: the
announcement. With even a short announcement of 5 seconds or
so, difficulties over pitch disappear. The ear does not register a
dissonance if two sounds are spaced and attention is distracted by
speech.
Adding speech to music
Where speech is added to music, as in the case of theme music with
announcements or introductory words over, the voice is normally
chosen for high intelligibility. The music can therefore often be
given full value and dipped just in time for the first words. Similarly,
ensuring that only the last word is clear, the music is lifted tight on
the tail of speech. Also, in anything longer than a breath pause in
speech, the music is peaked up. This is called a `newsreel' commentary and music mix after the old style of filmed newsreels in which
the inconvenience of synchronized speech, or even sound effects,
was avoided.
At the end of a programme, an upbeat finish may demand that music,
applause and speech all compete: here, similar techniques may again
be employed. But in neither case should it be forgotten that, for
many of the audience, disengagement is accelerated only by what at
home just sounds like a loud and not particularly pleasant noise. An
alternative, more relaxed mix has gentle music fades before and after
the words.
[14.7] Film music and commentary
Above: Normal cueing. Music (A) is
gradually adjusted in volume so that
commentary or other speech (B) is
clearly audible when it starts. After the
speech ends, the level is adjusted
gently. Alternatively, the music may be
written or chosen specially to give this
effect. Below: `Newsreel' cueing.
Music (A) is held at high volume until a
moment before speech (B) starts, and
is raised to high level again immediately
speech ends. All adjustments are rapid.
Chapter 15
Sound effects
Sound effects can reach the console either from a studio microphone
or from a replay machine. There are three main types: spot effects,
`library' recorded effects and actuality recorded effects.
Spot effects ± also called manual effects, hand effects or foley ± are
those that are created live. In a radio or television studio they are
generally produced at the same time as performers speak their lines.
In a film sound studio they are created to match action on the screen.
For the visual media, they are kept strictly off-camera.
`Foley', originally an American film term for synchronous spot
effects, is named after a 1950s practitioner who bridged a gap during
which the technique had gone out of fashion in Hollywood. In Britain,
where they were never out of use, the long-established name for them
was `spot', so that is what they are called in this book. Spot effects
include the noises of doors, telephones, bells, teacups, letters being
opened, crashes, bangs, squeaks, footsteps and even that old standby,
horses' hooves. They include noises it is simply not worth recording,
noises that are so much part of the action that to have them anywhere
but alongside it might leave an actor confused and uncertain of timing.
Recorded effects consist principally of those that cannot conveniently be created in the studio: cars, aircraft, birdsong, weather,
crowd chatter, and so on. Their field overlaps that of spot effects:
for instance, rather than those coconut shells, a recording of real
horses' hooves may be preferred ± although, curiously enough, some
very realistic horses' hoof sounds have actually been recorded with
coconut shells. It may seem odd that effects may be created by
apparently and in this case ludicrously inappropriate sound sources.
But it is important to realize that the purpose of sound effects is not,
in general, to recreate actuality, but to suggest it. Realism is not
always necessary, and at times may even be detrimental to the final
result, in that some inessential element of the `real' sound may
distract the listener.
240
Sound effects
Library effects are those already available on disc (usually CD), hard
disk or possibly tape; actuality effects are those specially recorded
for the occasion (more often for film or video than any other
medium).
For any radio production in which effects appear it is probable that
both spot and recorded effects will be needed, and a quick skim
through a script should be enough to see which category each cue
belongs to. The division of labour will often work out about fiftyfifty, with a few cases where the final decision depends on factors
such as whether the studio is equipped with the heavier pieces of
spot effects furniture or whether a particular recording can be found.
Several people might divide this work: the team leader at the console
is responsible for balance; one assistant looks after all recorded
effects including on-site editing and another creates effects in the
studio ± although the spot effects may be pre-recorded (often
directly to a digital editing system), then dropped in as required.
The conventional use of effects
Whether for film or television, and even more for radio, effects are
rarely used in a strictly realistic way. Even when that is attempted, it
is generally a heightened realism that owes a great deal to art.
As an example consider a radio report: an eyewitness account of a
`Viking Raid', re-enacted in a traditional festival. The tape received
in the studio could not be used as recorded. The reporter had, rightly,
concentrated on getting good speech quality, so that the background
noises were too low. There were some good shouts, screams and
crowd noise, but following the words `listen to the clash of their
swords', the swords were neither clear nor close enough. Also, there
were breaks in the recording, and whenever public address loudspeakers came on to explain the event to onlookers, the atmosphere
was destroyed.
To recreate the scene for a radio audience, the recording had first to
be edited into a compact, continuous narrative, deleting inessential
expressions of enthusiasm ± the report conveyed this without the
need for words. As reconstructed, the first sound heard was a
recorded effect, heavy breakers on the beach (the report referred to
gusts of wind up to 40 miles an hour), and these were held behind the
studio narrator, who sketched in background information. Toward
the end of this the actuality screams were faded in, and the taped
report took over. The effect of sea-wash was faded down as the
description of the invasion brought the `Vikings' on to the beach, but
an occasional squawk of seagull was dropped into the clamour to
keep the sense of location strong. After a first reference to swordplay
the words from `listen to . . . ' were cut, and in their place a closer
clash of swords was inserted. The final result was highly realistic: an
actual event (with the real shouts and screams) brought into focus by
a mix that included a little retouching by adding sounds that could
Sound effects
241
not have been recorded at the time without an array of microphones
all over the beach.
That is what might be called a neo-realist technique, well suited to
the sound medium in a programme that consists mainly of edited
actuality. But for a programme that is created entirely in the studio,
we must work within even more artificial ± but effective ± conventions; we rarely strive after the kind of sound that is found in a reallife recording.
Sound effects must often be held to an unrealistically low level: in
real life they may be just too loud. Another reason for this is that
masking caused by any background sound is worse in mono than in
stereo. The `cocktail party effect' is the ability of the ear and brain to
distinguish a single sound source in a crowded room, rejecting and
often almost ceasing to be aware of surrounding noise. When a
monophonic link (microphone, recorder or loudspeaker) is introduced into the chain, much of this ability to discriminate is lost. In
stereo the sound is clearer: it reduces the need for effects to be held
so far back from the rest of the action ± although unfortunately,
because it may still have to be compatible with mono, it may be
impossible to take full advantage of this. When adding stereo effects,
listen occasionally to the `M' signal to check the maximum acceptable level.
Loudness is nothing like so important as the character of a sound, the
evocative quality that can make it almost visual. This is true even of
effects for which loudness might seem an essential part ± for example, a car crash. Analysis of a typical car crash reveals a series of
separate components: skid, impact, crunching metal, and broken
glass falling to the ground. In a real crash the impact is the loudest
part: in fact, it is so loud that if it were peaked exactly to the loudest
permissible volume for no appreciable distortion, all the rest would
be far too quiet and lacking in dramatic effect. So the whole sound
has to be reorganized to ensure that no part is very much louder than
any other. Even more important, during this recomposition certain of
the characteristics have to be modified to create the right mood for
the sound to fit into its context. The comedy actor driving into a
brick wall makes a different noise from a payroll bandit doing
exactly the same thing. One must be ludicrous, the other vicious
and retributive.
The various stages of the crash may be made up as follows:
1. The skid ± essential to set up the action. A crash without a skid is
almost over before we can tell what is going on. The scream of
tyres is a distinctive noise that may not always happen in real life;
whether it happens in radio depends on mood. The skid is nemesis, doom closing in on the bandit.
2. A pause. The listener, having been delivered the initial shock,
must be given half a second to hold breath and . . . wait for it . . . .
3. Impact. A hard, solid thud ± difficult to get just right, but not so
important as the other sounds.
4. Crunching metal. In a real crash this can sound like a couple of
dustbins banged together, sufficiently ludicrous for the comedian,
242
Sound effects
but not sinister enough for the bandit. It needs to be purposeful
and solid, less like an empty tin can.
5. Glass. Like the skid, this is an identifying sound, and also one
that clearly puts an end to the sequence.
6. A period of silence ± doom has struck.
No sound effect should be just the noise that would naturally accompany the action. Rather, it must be related to dramatic content and
the intended emotional response.
Surrealistic effects
Effects that are intended to be funny are different from ordinary
dramatic effects. Action effects may be exaggerated in a subtle way,
by slightly emphasizing characteristic qualities and cutting away
inessentials, but in the comic effect emphasis of character is carried
to an illogical extreme. A whole canon of slapstick, or what in
Britain have been called `cod effects', is created by this method,
and in a fast-moving comedy show is deliberately overdone.
Consider an effect in the old Goon Show, where the script simply
said `door opens'. An ordinary door opening may or may not produce
a slight rattle as the handle turns, but for comedy this characteristic is
seized upon: if it were not for this, it would be hard to find any readily
identifiable element in the otherwise quiet sound ± a faint click would
be meaningless. For a `cod' door, the volume and duration of this
identifying characteristic are exaggerated beyond normal experience ±
though not so much as to obstruct the flow of dialogue.
Such a method of codding effects works only if the normal spot
effect makes use of some slightly unreal element: where the natural
sound cannot conveniently be exaggerated, other methods must be
found. For example, little can be done with the sound of `door shuts',
which in any case has to be toned down a little to prevent overmodulation. So other methods have to be examined: for example, the
creation of completely different and totally unreal sounds. These
might include such cod effects as a `swoosh' exit noise (created,
perhaps, by speeding up the change of pitch in a recording of a
passing jet plane). Or an object (or person) being thrown high into
the air might be accompanied by the glide up and down on a
swannee whistle (which has a slide piston to govern its pitch). The
villain in a comedy show may be dealt a sharp tap on the Chinese
block; or in a children's programme, Jack and Jill may come tumbling down a xylophone arpeggio.
As with dialogue, timing can help to make noises comic. This is
typified by the way many a crash ends. An example is Don Quixote
being thrown from his horse by a windmill: a whole pile of `armour'
lands on the ground, it is stirred around, followed by a pause. Then
one last separate piece of `armour' drops and rocks gently to rest. To
create this effect, a pile of bent metal discs was assembled and
threaded at intervals along a piece of string ± except for the last
Sound effects
243
single. Any similar pile of junk might be used. The timing of such a
sequence is its most vital quality; the essence of its humour could
almost be written out like music ± indeed, an ear for music is very
useful to an effects operator.
Normally a script should be concise, but just occasionally may take
off on a flight of fancy: one Goon Show script demanded `a 16-ton,
11¤2-horsepower, 6-litre, brassbound electric racing organ fitted with
a cardboard warhead'. Here anything but sheer surrealism was
doomed to stick on the starting line. The first stage in its creation
was to dub together a record of an electric organ playing a phrase of
music first at normal speed, then at double, and finally at quadruple
speed. Three spot effects enthusiasts dealt with gear changes and
backfires, engine noises and squeaks, plus hooters, whistle and tearing
canvas ± mostly produced by rather unlikely looking implements ±
and the whole effect was rounded off by a deep, rolling explosion.
Spot effects
A radio drama studio needs a handy supply of spot sound sources.
Some may be kept permanently in the studio; others might be collected from an Aladdin's cave of noisy delights ± a spot-effects store.
Studio equipment may include a spot door, a variety of floor surfaces, a flight of steps, a water tank and perhaps a car door. It is
difficult to find an adequate substitute for the latter, though a reasonable sound can be produced with a rattly music stand and a stout
cardboard box: place the stand across the box, and bang the two on
the floor together.
A well-filled store (sadly, now rare) can take on the appearance of
a classified arrangement of the contents of a well-appointed rubbish
tip. The shelves are loaded with stacks of apparent junk, ranging from
realistic effects such as telephones, clocks, swords, guns (locked in a
safe), teacups and glasses, door chimes and handbells ± none of which
pretended to be other than they are ± through musical sound sources
such as gongs, drums, blocks, glocks, tubular bells and the rest of the
percussion section ± which may appear as themselves or in some
representational capacity ± and on again to decidedly unobvious
effects, such as the cork-and-resin `creak' or the hinged slapstick
`whip'. In this latter field, ingenuity ± or a tortured imagination ±
pays dividends. One can, for example, spend days flapping about with
a tennis racket to which 30 or 40 strands of cloth have been tied (to
represent the beating wings of a bird) only to discover later that the
same sound could have been obtained with less effort by shaking a
parasol open and shut. A classic example is the French Revolution
guillotine: slice through a cabbage.
At one time, rain effects were produced by gently rolling lead shot
(or dried peas) around in a bass drum, sea-wash by rolling them
rather more swooshily, while breakers would be suggested by throwing the shot in the air and catching it in the drum at the peak of the
244
Sound effects
swoosh. This would now be used only for a deliberately stylized
effect; for a more realistic sound, a recording would be used. But a
bass drum is still a useful item in a spot store: a noise from something on top of it is endowed with a coloured resonance that can
overemphasize its character for comic effect. For example, brontosaurus' footsteps can be made by grinding coconut shells into blocks
of salt on a bass drum.
A vintage theatrical wind machine has a sheet of heavy canvas, fixed at
one end and weighted at the loose end, draped over a rotating, slatted
drum. This would now be used only as a deliberately theatrical effect
like that in an orchestral work such as Ravel's Daphnis and ChloeÈ,
where the ability to control pitch is valuable. But realistic wind comes
in many qualities, and its tone should suit the dramatic context. A good
recorded effects library has gentle breezes, deep-toned wind, highpitched wind, wind and rain, wind in trees, wind whistling through
cracks, and blizzards for the top of Mount Everest or the Antarctic.
A wind machine produces just one sound ± wind; the same applies to
thunder sheets. They are not convincing except for comic effects.
But whether or not such ingenious improvisations have a certain
period flavour, there are certain basic qualities of spot effects that
are as valid today as they were 50 years ago. With high-quality audio
transmission, the acoustic requirements for effects are now more
stringent than in the distant past. The need for good timing, however,
does not change: it is still the most important quality of a wellperformed spot effect ± all the more so because it is likely to be a
staccato interjection, while continuous background sounds are more
often supplied in the form of recordings.
Where feasible, the radio spot effects operator works alongside or
behind the performer, their relative positions depending on the
stereo image: this helps to match acoustic and perspective, as well
as timing. Even large pieces of equipment can be set in the correct
perspective. Where this proves impossible ± for example, for footsteps that are restricted to an area where a suitable surface is
provided, or for water effects that depend on the use of a tank ±
the effects must be panned to their correct position, and if necessary
given stereo spread by using a pair of microphones, or perspective
by adding stereo artificial reverberation (AR).
If a separate microphone is necessary it will often be hypercardioid,
for best separation. Where the same studio is used as that for action
and dialogue, it is usually fairly easy to match the acoustics. Matching
that of film has sometimes required more effort: a studio effect can
sound too `clean'. Here AR with a large number of pre-set acoustics
may be the solution, and atmosphere of some kind will be added.
Doors
In a studio used for drama, the most obvious item of studio spoteffects equipment is a large square box on castors, generally with a
Sound effects
245
full-size door set in it. The handle is a little loose, so it can be rattled
as it opens. Other door furniture might include a lifting latch, bolt,
knocker, etc. Also helpful would be a sash window frame (unglazed)
fitted in the back of the box. If no spot door is available, do not
despair: many existing effects recordings were made with equipment
like this, and the principles can still be employed.
The sound of opening has to be exaggerated somewhat: a more
natural version may be too quiet to register. The sound of closing
may have to be softened a little to avoid overmodulation. Close
the door lightly but firmly. The characteristic sound is a quick
double-click.
Bad timing will impede the flow of a programme. The sound of a
door opening should generally come slightly ahead of the point at
which it may be marked in a script; this allows dialogue to carry on
without any pause. If a scripted cue follows a completed spoken line
of dialogue, superimpose it over the last word, unless there is reason
to leave that word cold (e.g. to follow a direction to enter). If the
door is to be closed again, allow sufficient time for the actor to `get
through' it. Also, the perspective of actor and door should match,
and they should stay matched until the door is shut again, if that is
part of the action.
[15.1] Sound effects door
This is used in a radio drama studio.
Some types of door always have to be pre-recorded: prison cell
doors, castle gates, and so on. Sometimes heavy doors of this kind
can be made up from the sound of an ordinary door by transposing it
to a deeper pitch and adding AR. An older method was to replay a
recorded door at reduced speed. Halving the pitch or halving the
recorded speed doubles the size and weight of the door, but does not
change the characteristic resonance of a material such as wood.
When knocking, take care to find a place where the door or frame
sounds solid enough. Metal knockers, bolts, latches, lock and keys,
and so on, should be heavy and old-fashioned; they are best attached
to the effects door or, if loose, held firmly against it. The larger area
of wood acts as a resonator and gives the sound a truly solid quality.
Loose bolts, etc., should be screwed to a small panel of wood.
One last note about doors: when in doubt, cut. Some scriptwriters
overdo the mechanics of exits and entrances. An approach on dialogue is often as effective as door-plus-approach. It is better to keep
doors, like all other effects, for the occasions when the sound has
some real significance in the story. Otherwise, when the time comes
that it does, it will not be so easy to establish its importance.
Footsteps
What was a footnote on doors becomes a point of primary importance with footsteps. Are your footsteps really necessary? Go through
a script with a `delete' pencil and make sure that every remaining
footstep has direct relevance to the action. In films this effect is often
246
Sound effects
used merely to fill a silence, to reassure the audience that the sound
has not broken down at a time when all attention must be concentrated on the picture. There is no direct equivalent to this in the pure
sound medium, in which footsteps can all too easily draw the
listener's attention in the wrong direction.
Such footsteps as remain may be created as spot effects or played in
from a recording: the former is preferable, because the sound effect
should tie in exactly with the mood of the moment, while also
reflecting character. In the studio the exact pace can readily be set
and changes of pace and a natural stop produced exactly on cue.
A well-equipped radio drama studio or film sound studio should
have a variety of surfaces available for footsteps: the most important
are wood, stone and gravel. Boards or stone slabs (laid on top of the
studio carpet) should be as hefty as is convenient, to give weight to
their resonance: the carpet stops them ringing too much. A gravel
effect can be created by spreading sand and gravel on the stone slabs.
[15.2] Gravel `pit' for footstep
effects
This pit is built into the studio, and
concealed below the wooden floor.
On these surfaces long sequences of footsteps are difficult. Walking
on the spot is not really a satisfactory way to create footsteps, as it
takes an expert to make them convincing. Walking along, the foot
falls: heel, sole, then heel, sole again. When walking on the spot it is
natural to put the foot down sole first, so a special heel-first technique must be worked out: a fairly crisp heel sound, followed by a soft
sole sound. This is not at all easy to get just right. Even when the
foot action is the same, the two are still readily distinguishable by
the fact that the surface `ring' sounds different on every step when
walking along, but is monotonously the same when the foot keeps
landing at the same place.
Walking or running up and down stairs are also better done with real
steps: there can be a very different quality for each step. But if the
effect has to be simulated, the correct footfall is with the sole of the
foot, going up, or sole±heel, going down. Again, the result is difficult to get right unless there is some variety of surface.
Attention should be given to footwear. Leather heels and soles are
best for men's shoes. High-heeled women's shoes are usually noisy
enough without special care in their choice. If only one person is
available for footstep effects, remember that it is easier for a woman
to simulate a man's footsteps than vice versa. It is not easy to get the
right weight of step if shoes are held in the hand. The simplest way
of getting two sets of footsteps is to have two people making the
sound: except by pre-recording, one person cannot conveniently
do the job, especially as the two are not generally in step. But two
sets of footsteps will stand for three, and three sound like a small
crowd.
Some surfaces are simulated. For footsteps in dry snow, cornflour
(in America, corn starch) has been used, but this can be messy.
A fine canvas bag of cornflour can be crunched between hands,
or a spoon (or two) pressed into a tray of the powder. For a frosty
path try salt; for dry leaves or undergrowth, unrolled reel-to-reel
tape.
Sound effects
247
Telephones, bells and buzzers
For electric bells and buzzers it is useful to make up a small batteryoperated box that can be used to give a variety of sounds (or make up
several boxes, each with a different quality). The mobility of the box
allows it to be placed at any appropriate distance, in any suitable
acoustic, or against any surface that will act as its sounding board.
For telephones in radio, a standard handset can be adapted, so that
the ringing tone may be operated by a press-button. In order to
suggest that picking up the hand-piece stops the telephone ringing,
try to finish up with half a ring. For example, a common ringing tone
in Britain is a repeated double buzz-buzz . . . buzz-buzz; the ideal
interrupted ring goes buzz-buzz . . . buzz-b . . . followed quickly by
the sound of the handset being lifted. As this last sound is much
quieter than the ringing, two things can be done to make it clear.
First, the rattle of picking it up should be emphasized a little, and
second, the telephone can be brought closer to the microphone at the
end of the effect. The first ring is fairly distant (perhaps partly on the
dead side of a directional microphone). This is doubly convenient in
that it may be interrupting speech, for which purpose it is better not
to have a foreground effect. The second ring is closer, as though
a camera were tracking in to the phone, to become very close as
it is picked up. The words following must also be in close perspective.
The action of replacing the receiver may be balanced at a greater
distance. Naturally louder than the pick-up, it is an expected sound
that fits neatly in the slight pause at the end of a conversation.
Dialling is more time-consuming than being on the receiving end of
a call, simply because most codes are much too long: 10 or 11 digits
are commonplace. If there is any way of cutting this sort of hold-up
from the script it should usually be adopted, unless dialogue can
continue over dialling. Period rotary dials are a menace: codes were
shorter, but even so, it is better to dial only a few numbers.
For television, telephones are used so often that it is worth making
up `specials' ± a portable kit using real telephones, properly (and
permanently) polarized so that they are fully practical, and linked
together. This is necessary because good sound separation between
the microphones picking up the two halves of the conversation may
make it difficult for the parties to hear each other clearly without the
telephone. If one of the speakers does not appear in vision it is better
to use an acoustically separate booth.
[15.3] Effects bell
This home-made device has two
press-button controls.
If both speakers do appear in vision at different times their voices are
balanced on normal studio microphones, but that of the speaker who
is out of vision must be heard through a telephone-quality filter.
Manual switching is risky, as the result of the slightest error may be
obvious. A better arrangement is therefore for the filter circuit to be
linked to camera cut buttons so that the two ends are automatically
reversed on a cut. When there are more than two cameras (i.e. more
than one at each end) it is necessary to make sure that any camera
248
Sound effects
that can possibly be used in the sequence is linked in to the switching
system.
[15.4] Control unit for telephones
used in television plays
1, Input sockets for up to six lines.
2, Keys to select mode of operation
(manual or automatic) for each line.
3, Line keys: down to select, up to
cancel. 4, Indicator lamps to show
line `called'. 5, Indicator lamps: `line
calling'. 6, Ringing switch for use by
operator: up for automatic ringing;
down for manual ringing. 7, `Autodial' switch: when `on', this permits
extensions with dialling facilities to
ring each other automatically; in
addition, normal dialling tone is heard
when a receiver is raised. In all other
operations the switch is left `off'.
8, Ringing frequency selection.
A simpler arrangement for
interconnecting telephones
is often adequate.
A kit can be set up with a number of telephones connected through a
box that acts as a small telephone exchange that can produce manual
or automatic ringing tones at a range of audio frequencies. Lights
above each key show the condition, `called' or `calling', of the
associated line. When the automatic button is pressed the selected
telephone rings immediately, beginning with the start of the first
buzz, and it goes on automatically until either the operator's finger
or the phone is lifted, when the ringing stops in a realistic manner.
Using automatic telephones also ensures that the bell is in the same
acoustic and perspective as the actor. Take care to use appropriate
ringing tones: different countries have their own standard signals.
Most `far end' dialling, ringing or engaged tone effects can be
provided by using a buzzer, etc., on strong filter or by playing a
recording.
Personal action effects
There are many sounds that are so individual in their application that
it would be almost ludicrous to use recordings: sounds like a newspaper being folded, opening a letter, undoing a parcel, or eating and
drinking.
First, paper sounds. Newspaper makes a characteristic noise, but in a
spot effect the action may be formalized somewhat, e.g. shake
slightly, fold once and smooth the page. Opening a letter should be
similarly organized into two or three quick gestures, the last action,
smoothing out the paper, being tucked under the words that follow.
Work fairly close to the microphone. The paper should be thin but
hard surfaced: brown paper can crackle like pistol shots, and airmail
paper sizzles like a soda siphon. Keep the actions simple but definite. For undoing a parcel: a cut, perhaps, then two deft movements to
smooth away the paper, and a final slower crinkling noise trailing
under speech.
Many sounds, such as an orchestral conductor tapping his baton, can
be made realistically (using a music stand); similarly for a judge's
gavel and block (but soften this just a little to avoid sharp peaks, by
binding a layer or two of electrician's tape round the face of the
gavel). The surface struck must have the proper resonance: it will
not do to hold a small block of wood in the air; it should be on
something solid.
Pouring a drink is also a spot effect. The sound of pouring may not in
itself be sufficiently definite, and for all but the longest of drinks is
over too quickly; so a clink of bottle on glass can help. Even without
the clink it is important to use the right thickness of glass. Effervescent health salts do fine for fizzy drinks. A popgun may be a shade
less acceptable as a champagne cork than the real thing, but is more
Sound effects
249
predictable. If `tea' or `coffee' is to be poured into a cup, the effect
can be confirmed by placing a spoon in the saucer afterwards ± and
again the right thickness of china is essential.
For a scene set at a meal table, the complete spot kit includes knife,
fork, spoon and plate, plus cup and saucer. But give the noise a rest
from time to time: the right points to choose for action depend on
mood and pace of the script. At any point where the tension rises,
stop the effect; as the mood relaxes again, the knives and forks
resume. It takes only a little of this noise to suggest a lot: listen to
a recording of a tea party for six and you will find that it sounds like
a banquet for 20.
Gunplay
A rifle shot is one of the simplest sounds you can meet. It is an
N-shaped wave like that from a supersonic aircraft: a sharp rise in
pressure is followed by a relatively gradual drop to as much below
normal pressure. Then there is a second sharp rise in pressure, this
time returning it to normal, then nothing ± nothing of the original
sound, that is. The length of the N (its duration in time) depends on
how long the object takes to pass a single point. For a supersonic
airliner this is short enough ± for a bullet it is almost instantaneous:
there is just amplitude and brevity. A rifle shot travels at something
like twice the speed of sound, and that from a smaller gun at close to
the speed of sound. Whatever the speed, any loudspeaker would
have a hard job to match the movement, however simple. It is
fortunate that what happens after the double pulse is more complex.
In fact, the very size of the initial pulse is, both in recording or
broadcasting, totally unacceptable; it is necessary to discriminate
against it. One way is to fire (or simulate) the shot on the dead side
of a microphone; another is to take it outside the studio door, or into
some small room built off the studio, so its door can be used as a
`fader'. Very often a script calls for loud shouts and noisy action
when shots are fired, which makes it easier to deal with the sheer
loudness of the effect than if the microphone has to be faded right up
for quiet speech. The best balance varies from case to case, so
experiment is always necessary, but the shot can be as loud as the
equipment can comfortably handle. Check by ear for distortion: the
full peak may not show on a meter. Perhaps use a limiter, with rapid
recovery, and a second, distant microphone angled to pick up
reverberation only.
Experiments are also needed to select the appropriate acoustic: the
sound picked up is a picture of the acoustics in their purest form.
Shots recorded out of doors are all bang and no die-away, which
means that at the level at which they must be replayed they can
sound like a firecracker. Always check the acoustic when balancing
a studio shot, and watch out for coloration and flutter echoes that
may not be noticeable on a more commonplace sound.
250
Sound effects
A third element in the sound is (or may be) a ricochet ± the whining
noise made by the jagged, flattened, spinning scrap of metal as it
flies off some solid object. And whereas the previous sound is just a
noise of the purest type, a ricochet is a highly identifiable and
characteristic sound. Even if in reality ricochets are less common
than they might appear from Westerns, they may be used in practically any outdoor context, i.e. anywhere there is enough space for
a ricochet.
Some of the techniques that have been used for gun effects are
outlined below:
.
.
[15.5] Gunshot and other
percussive effects
Above: Slapstick. Centre: Wooden
board and hard floor surface. Below:
Gun modified to prevent firing of live
rounds.
.
.
.
.
.
A real gun. A modified revolver is better than a starting pistol,
which sometimes tends to sound apologetic. Legal regulations
must be properly observed and, needless to say, care used in
handling the gun ± even blanks can inflict burns. Some guns for
radio use are modified to allow the gas to escape from a point
along the bottom of the barrel, not from the end (so that it is not
usable as a real gun). The microphone balance adopted is generally that for a high ratio of indirect to direct sound.
An electronic gunshot generator or a hard disk variation on this.
This is essentially a white noise generator that when triggered
produces `bangs' with a variety of decay characteristics, ranging
from staccato to reverberant. There may also be associated circuitry (or a program) for producing a range of ricochets.
A long, fairly narrow piece of wood, with a rope attached to one
end. Put the wood on the floor with a foot placed heavily on one
end, then lift the other end off the ground with the rope and let go.
The quality of the sound depends more on the floor surface than it
does on the type of wood used. A solid floor is best ± stone or
concrete.
Cane and chair seat. The sound quality again depends on the
surface struck. Padded leather gives a good crisp sound.
A slapstick. This is a flat stick with a hinged flap to clap against
it. Depending on balance and acoustic conditions, it may sound
like a gunshot (though not a good one), the crack of a whip, or
just two flat pieces of wood being clapped together ± literally,
slapstick! Narrow strips of thick plywood are as good as anything
for this.
A piece of parchment on a frame, with an elastic band round it ±
the band can be snapped sharply against the resonant surface.
Again, this is more slapstick than even remotely realistic.
Recordings of real bangs or of any of the above. The apparent
size of any explosion can be increased by slowing down a
recording ± a sharp clap may become a big, rumbling explosion
if copied down to an eighth or sixteenth of the original speed.
Alternatively, reduce pitch and increase duration independently
by digital signal processing.
If the gun itself is used in radio it can be taken well away from the
action. For television this may not be possible: particularly if the gun
is to be fired in vision, perhaps several times, with realistic flashes
and smoke and perhaps with dialogue before, between and after. In
this case, again, use two microphones. On the close (dialogue)
Sound effects
251
microphone set a limiter for 2 dB below full modulation; make the
second distant, for a high proportion of indirect sound, perhaps in
this relatively dead acoustic with added AR.
For television gunshots out-of-vision or for radio the simplest playin arrangement is to set up a MIDI keyboard to trigger a variety of
gunshot effects. One possibility is to link this directly to a studio
loudspeaker that is balanced acoustically. Then, as they can hear the
bangs, the actors can react to them directly.
The original sound might again be created by a white noise generator. This passes through a gate (operated by the `trigger', however
this is keyed) and a low-pass filter. The gating circuit allows the
white noise passing through to rise sharply from zero to a predetermined peak, automatically followed by an exponential fall that
eventually shuts off the white noise once again. With a low-pass
filter set for 5 kHz cut-off it sounds much like the close shots of a
revolver or automatic; if the filter is set to remove all but the lowest
frequencies, it gives distant heavy gunfire ± a `cannon' quality. For a
machine-gun effect a device (or program) operates the gate at
regular intervals for as long as the key is pressed. In this case the
decay is made steeper in order to keep the shots separate and
distinct. Ricochets require a separate pulsed output that is rich in
harmonics and falls in frequency with time. This might work from
the same initial trigger action, but delayed a little.
[15.6] Simple gunshot effects
generator: general principles
1, White noise generator. 2, Push
button. 3, Impulser relay. 4, Gate.
5, Low-pass filter. 6, Studio
loudspeaker.
Creaks, squeaks, swishes,
crackles, crashes and splashes
Creaks and squeaks offer rich opportunities for improvisation. Given
a little time, the enterprising collector can assemble a vast array of
squeaky and creaky junk just by noticing noises as they happen and
claiming the object that makes it. Wooden ironing boards, old shoes,
metal paper punches, small cane baskets; anything goes ± or stays.
Two well-established ways of producing creaks and squeaks are:
.
.
String, powered resin and cloth. The resin is spread in the cloth,
which is then pulled along the taut string. If the string is attached
to a resonant wooden panel the sound becomes a creaky hinge.
Varying the pressure on the string changes the quality of sound.
A cork, powdered resin and a hard surface, e.g. polished wood, a
tile or a saucer. Place some resin on the surface and slowly grind
the flat of the cork into it. Again, the squeakiness varies with
pressure, speed, type of surface and any resonator that it is in
contact with.
A bright, crackling, spitting fire may be simulated by gently rolling a
ball of crinkled cellophane between the hands.
[15.7] Creaks and squeaks
Methods using resin and string;
or cork, resin and tile.
For the flight of a javelin or arrow, swish a light cane past the
microphone. Here realism is not required: in real battles the arrows
252
Sound effects
bounced off stone walls and parapets, making an undramatic clatter.
An arrow landing in wood sounds far more satisfying, so convention
demands that all the best misses land in wooden panelling and treetrunks. For this, throw a dart into a piece of wood near a microphone.
Proximity makes a sound grow bigger ± so the twang of an elastic
band may do well enough for the bow.
Crashes are another reason for hoarding junk, ranging from empty
cans to pretty well anything that makes a resonant noise when
dropped. But real glass crashes demand care. For these, spread out
a heavy ground sheet and place a stout cardboard box in the middle.
Sprinkle some broken glass in the bottom and place a sheet of glass
across the open top of the box. Wearing glasses or goggles and
gloves for safety, give this sheet a sharp tap with a hammer. As it
shatters it should all fall inside the box. Have a few odd pieces of
glass in the free (gloved) hand to sprinkle in just afterwards. Not
difficult ± but pre-record it, to avoid the danger of having broken
glass around on the take (the same goes for footsteps in broken
glass). Another reason for pre-recording is that successive crashes
are not predictable in quality or volume.
One of the fittings for a well-equipped drama studio is a water tank.
For convincing water effects, the tank must be nearly full, to avoid
any resonance in the space at the top. For any but the gentlest water
effects, protect the microphone ± which may have to be close ± by
hanging a gauze in front of it, or putting it in a plastic bag.
Horses' hooves
Horses' hooves are included here as an example of a specialized
spot-effects technique that may seem dated, but is still valid. Coconut shells really can be as good as the real thing, if not better ± and
are certainly a lot easier to fit with dialogue.
The secret of a good hoof sound lies not so much how to make the
shells ± take one coconut, saw it neatly across the middle, and eat the
contents ± as in creating the surface to use them on. Start with a tray
about 1 m (3 ft) across and perhaps 10 cm (4 in) deep (e.g. a baker's
tray). Fill the bottom with a hard core of stones and put gravel and a
mixture of sand and gravel on top. For a soft surface, place a piece of
cloth or felt on top of this at one end of the tray, and bury a few half
bricks at the other end to represent cobbles. This will give the
qualities of surface ordinarily needed for horses' hooves. For horses'
hooves in snow, fill a tray with broken blocks of salt.
[15.8] Horses' hooves
Traditional method: a pair of half
coconut shells.
There are three basic actions to simulate: walk, canter and gallop.
The rhythm is different for each. When a horse walks, the rhythm
goes clip, clop . . . clip, clop . . . . With a shell in each hand, use one
then the other, but with alternate steps from each shell move forward,
digging in the `toe' of the shell, then back, digging in the `heel'. The
sequence is left forward, right back (pause), left back, right forward
(pause) . . . . In this way a four-legged rhythm is obtained. For the
Sound effects
253
canter, the same basic backward and forward action is used, but the
rhythm is altered by cutting the pauses to give a more even clip, clop,
clip, clop, clip, clop, clip, clop. In galloping, the rhythm alters to
a rapid clipetty clop . . . clipetty clop . . . 1, 2, 3, 4 . . . 1, 2, 3, 4 . . . . In
this case there is hardly time for any forward and backward
movements.
For a jingling harness, use a bunch of keys and a length of elastic:
hang the keys on the elastic between a little finger and a solid
support to one side, so the keys jingle in rhythm with the horse's
movement.
A separate spot-effects microphone is needed for hooves, usually a
hypercardioid. This should not be too close, and the acoustic should
match that used for dialogue.
The recorded sound picture
There are two distinct types of actuality sound recording: some
create a sound picture; others are selective of a particular sound
element to the virtual exclusion of all others. These two categories
are different not only in content, but are also based on totally
different ideas about how sound effects might be used. The sound
picture is the complete picture, a story in its own right. But a
recorded sound effect, like the spot effect, should be heightened
reality: the distilled, essential quality of location or action. It is a
simplified, conventional sound, the most significant single element
of a complete picture.
A sound picture of a quiet summer countryside may have many
elements, with a variety of birdsong, perhaps lowing cattle a field
or two away, and in the distance a train whistle. A beach scene may
have voices (shouts and laughter), the squawk of seagulls, a distant
rush of surf on shingle and, beyond that, the popping exhaust of a
motorboat. As pictures in their own right these may be delightfully
evocative, but how would they fit into a play?
It is easy to imagine dialogue to match such a scene ± but, significantly, the speech would have to fit to the sound effects, not the
other way round: in the country picture the cows might play havoc
with an idyllic love scene, and the distant train whistle would add a
distinctly false note to Dick Turpin's ride to York. Similarly, the
beach scene is fine ± if we happen to be on the sand and not in the
boat. And suppose we want to move about within this field of
action? With a sound picture we are severely restricted: the only
view we ever have is that seen by the artist from the point where the
canvas was set up; we cannot stroll forward and examine one or
another subject in greater detail unless this action is chosen for us.
But if we ourselves are given paints and canvas we can recompose
the scene as we wish. The true sound effect is the pigment, not the
painting.
254
Sound effects
In any programme where recorded effects are to be combined with
speech, the final composition will at many points be markedly
different from that of a sound picture that has no speech. In the
beach scene, the first sound heard might be the surge of surf on
shingle, and, as this dies, the hoarse cry of a gull; then, more
distantly, a shout and answering laughter. The second wave may
not sound so close, and from behind it the distant exhaust note of the
motorboat emerges for a few moments. The first words of scripted
dialogue are superimposed at a low level and gradually faded up to
full. At this stage the composition of the effects might change.
Staccato effects, such as gulls and shouting, should be more distant
but timed with the speech; brought up highest when the dialogue is
at its lowest tension, characters move off-microphone or in gaps in
dialogue (other than those for dramatic pauses). The motorboat, too,
may go completely or be missing for long stretches, appearing only
as a suggestion in a gap. The separate elements can be peaked up
more readily than the waves, which cannot be subjected to sudden
fluctuations in volume without good reason.
When using effects like this, avoid any that have become radio
clicheÂs. On British radio, because the long-running `Desert Island
Discs' starts with seagulls, sea-wash and a record of `The Sleepy
Lagoon', it has been difficult to use seagulls and sea-wash together
in that way. Be careful, too, about authenticity in details like the
regional location of birds and railways. There are so many experts
on these sounds, we should try to get them right ± and with birds the
time of year as well.
The use of recorded effects
As the aims differ when recording a sound picture or sound effects,
so also must the recording techniques. For sound effects, their backgrounds should not be obtrusive, nor their acoustics assertive. Continuous effects should have no sudden peaks or unexpected changes
of character.
The best microphone placing for an effects recording may not be the
most obvious. For example, a microphone placed inside a closed
modern car might not pick up any clearly recognizable car sound, so
for a car recording ± even for an interior effect ± the microphone
should be placed to emphasize continuous engine noise. But the
mechanical noises associated with gear changes should not be
emphasized (except for starting and stopping), as their staccato quality may interfere with the flow of the script.
An effects sequence, like the composite sound picture, is best constructed from several distinct components. Take, for example, a
sequence (to be used with dialogue) in which a car stops briefly to
pick up a passenger. The elements of this might be as follows:
1. Car, constant running.
2. Car slows to stop.
Sound effects
255
3. Tick over.
4. Door opens and shuts.
5. Car revs up and departs.
Of course, a recording might be found with all of these with the right
timing and in the desired order. But:
.
.
.
.
.
.
If there is much dialogue before the cue to slow down, the timing
is more flexible if the slowdown is on a separate recording. The
cue to mix from one to the next should be at a point in the
dialogue about 10 seconds or so before the stop, or there will
be too big a change in engine quality. As the sound of a car
stopping lacks identifiable character, a slight brake squeal or the
crunch of tyre on gravel might help to clarify the image. The
overall level of effects may be lifted a little as the car comes to
a halt.
The tickover might possibly be part of the same recording as the
stop. But the microphone balance for a clear tickover is so
different that it is probably better to record it separately. Fade
in the tickover just before the car comes to rest.
Doors opening and slamming must be timed to fit the dialogue.
A real car door shutting is loud compared with the door opening
and tickover, so it needs separate recording and control.
Departure could be on the same recording as the tickover, but
again timing is more flexible if the two are separate and the
departure can be cued precisely.
Extraneous noises, such as shuffling on seats, footsteps, etc., may
be omitted.
Additional sounds ± e.g. another door, a motor horn or the setting
of a mechanical taximeter ± can be cued in at will.
The best way of putting together sequences like this depends on the
replay facilities and sources available. Some constituent effects may
originate from hard disk libraries on the Internet, from which items
can be downloaded as part of a service for a fee and are then retained
on hard disk or another medium such as CD-R. Many users have
a more or less limited collection of commercially available CDs
(any one of these contains a large variety of effects, each on a separate band) and may buy an additional CD as required. Many make
their own recordings, then keep them for possible reuse. The essentials
for use are flexibility and speed of operation; it should be necessary
to pre-record only the most complex sequences. Hard disk can also
be used for shorter sounds, in some cases replacing spot effects, as
well as for pre-recorded announcements, jingles and commercials,
and for game shows where a limited number of stereotyped musical
effects are punched in to signal success, failure or as `stings' to mark
a change of direction.
By far the most flexible method is to transfer each effect to hard disk
for cueing via the keys of a MIDI controller ± a keyboard for greatest
versatility, or (for just a few cues) a picture of the keyboard on a
computer screen with a mouse to `press' the keys. Some can be
copied to nearby keys with slightly changed pitch: for example, the
same train door slamming twice becomes two different doors (also at
256
Sound effects
different pre-set volumes) as two adjacent keys are hit. A whole
sequence can be constructed with overlaps (layered effects) on a
sequence of keys with complete flexibility of timing. A performer
requires headphones to interact with it. For stereotyped effects and
`canned' applause, too, MIDI provides the perfect can. MIDI is
described further in Chapter 16.
A CD or hard disk library of effects can also be used with digital
editing (see Chapter 18) or for video soundtracks (Chapter 19).
Recorded effects in stereo
For stereo drama (whether for radio, television or film), it is best to
use an effect actually recorded in stereo, steering it to its appropriate
position and width. Where no suitable stereo recording is available
and it is impracticable to make one specially, it may be possible to
adapt existing monophonic sounds.
With continuous noises such as rain or crowd effects, a stereo picture
can be created by using several recordings played out of sync (or
one, with different delay time settings) and steered to different
positions within the scene. Rain or crowd scenes could also be fed
through a device called a spreader, which distributes a mono sound
over the whole sound stage like wallpaper.
[15.9] Spreader circuit
The monophonic input (M) is split
equally between paths X and Y. 1 and
2, Delay networks producing
differential phase shift. 3, Spread
control fader. 4, Sum and difference
network producing A and B
components from the X and Y
signals.
For a distant, moving sound in the open air (or one that passes
quickly), panning a mono source often works well enough. A larger
object such as a troop of horses passing nearby would require a
spread source to be panned across, with the spread broadened as the
troop approaches and narrowed again as it departs. This effect could
also be achieved by panning two mono troops across one after the
other, while a coach and four might benefit by spread horses from
one recording followed closely by wheels from another.
In an example illustrating all of these, a horse and carriage may be
heard to pan through a spread crowd, until a shot is heard from a
mono point, its reverberation echoing over the whole of the sound
stage. Note that in stereo radio drama there is a danger of lingering
too long on movement that is meaningless in mono: for the sake of
compatibility this should be avoided.
Changing the pitch of an effect
The range of sound compositions available from a given set of
effects recordings can be increased if their pitch can be changed ±
ideally, independently of duration. The digital method is to use a
digital signal processor (DSP) that includes a pitch- and time-shift
program. The older technique was to change the speed of replay of
the recorded effect (shifting pitch and duration together), and this
Sound effects
257
can also be achieved with a MIDI keyboard. Equipment designed
primarily to operate at given fixed speeds sometimes has a control
that allows continuously variable speeds between, say, half and
double the nominal speed.
When the replay speed of an effect is changed (so that pitch and
duration are shifted together), the most obvious difference in quality
is an apparent change of size: slow down the two-stroke engine of a
motor lawnmower and it becomes the lazy chugging of a motorboat,
or a small motorboat becomes a heavy canal barge. In the same way,
water effects can be increased in scale, just as filming a scaled-down
model boat in slow motion may simulate the real thing. Speeding up
an effect (raising its pitch and shortening the duration) gives the
opposite kind of change ± for example, mice talking. For out-of-theordinary effects these techniques can sometimes save a lot of work.
Shifts in pitch and duration also change the apparent volume of the
room in which an effect was recorded: the characteristic colorations
go up or down in pitch with everything else. This means that
recordings made out of doors may be more conveniently speedchanged than those with indoor acoustics attached. A good subject
for drastic transformation is wind. Record a good deep-toned wind
and double its speed to get a high, stormy wind; double again for a
biting, screaming Antarctic blizzard.
The recorded effects library
The contents of an effects library fall into a variety of categories.
First, and most important, there are crowds ± a crowd being anything
from about six upward. It may seem odd that the first requirement
among recorded effects should be the human voice in quantity, but
plainly, the illusion of numbers, the cast of thousands, is often essential if a programme is to be taken out of the studio and set in the world
at large. There are many subheadings to `crowds': chatter, English and
foreign, indoor and outdoor; the sounds of apprehension, fear, agitation, surprise or terror; people shouting, whispering, rioting, laughing,
booing and applauding. It may even just stand there and listen: the
sound of a crowd that is silent except for an occasional rustle, shuffle
or stifled cough can at times be as important as the roar of a football
crowd. A further subdivision of crowd effects is footsteps, in this case
the crowd being anything from two pairs of feet upward.
Transport effects include cars, trains, aircraft, boats and all the
others (such as horses) doing all the various starting, carrying on
and stopping things we expect of them, and recorded both with the
microphone stationary and moving with the vehicle, and with interior and exterior acoustics where these are different.
Another important group that has links with both of the above
is atmosphere of location, or `atmos' recordings. Streets and
traffic (with and without motor horns), train stations, airports and
258
Sound effects
harbours ± all these are needed, as well as the specific identifying
noises of London, Paris, New York and other major cities; also the
atmosphere of schoolroom, courtroom, museum, shop or concert
hall, and the distinctive sounds of a wide range of environments.
The elements ± wind, weather, water and fire ± provide another vital
field. Then there are the sounds of all branches of human activity:
industrial sounds, sport, domestic effects from the ticking of clocks
to washing up, and the sounds of war. There are animals, birds,
babies; and dozens of other classifications of sound, such as bells,
sirens, atmospherics, electrical sounds, etc.
A good recorded effects library (on whatever medium) should include
copies of all these and many more. They may be purchased from a
range of commercial sources, and broadcasting organizations, specialist production companies or individual enthusiasts may even add their
own. In a large library catalogue it should be easy to find and listen to
a dozen different recorded effects that are nominally suitable, to find
out which is best and which will not do at all. Typically, it is possible
to select two or three specimens of each sound: a likely best bet, plus
others sufficiently different in character to act as alternatives in case,
on adding sound to dialogue, the first turns out to be wrong in mood,
level or timing. A different recording can be readily substituted at any
point in a production. But skill in cueing (involving the judgement of
timing) is needed for programmes with complex effects.
Acoustic foldback
Pre-recorded sound effects are sometimes fed through a studio via a
loudspeaker and microphone. It is of greater advantage to do this
when the original recording is acoustically `dry' ± from the open air,
perhaps, or made with a close balance. Passing the sound through the
studio adds the same recognizable acoustic qualities that are heard on
the voices. In addition, the actors have something to work to, and can
pitch their voices against it in a realistic manner. A danger is that the
combination of loudspeaker and microphone can introduce coloration.
[15.10] Acoustic foldback
Reproducers (1), normally fed direct
to the mixer via switch contact (2),
may be switched (3) to provide
an acoustic feed to the studio
loudspeaker (4).
The volume of the studio loudspeaker amplifier should be set at a
level such that most of the range of the console `send' control is in
use. These two controls are used in combination to set the sound
level in the studio: experiment to find the optimum settings for the
various elements in the chain ± if a satisfactory result can be
obtained at all. This technique will not work if the effect is already
brighter than it should be and the studio is dead.
Effects for television and film
Most of the effects described above work well out-of-vision on
television. Recorded effects are likely to be preferred to spot effects,
Sound effects
259
which in turn are more often recorded in advance and played in: this
helps to limit the number of open microphones and produces more
predictable results when the director has so much to watch and listen
to at the same time. However, in-vision sound effects are still produced by some practical props. These need the same care in balancing
that they would have in radio, but in more adverse conditions.
Often an effect that can be picked up satisfactorily on a speech
microphone sounds much better if given its own closer effects
microphone. For example, a discussion about radioactivity using a
Geiger counter could have a small electret hidden near the loudspeaker of the counter. For this, a recorded effect could not be used
successfully because the rates would not vary realistically; because
the people present would not be able to react properly to them; and
because a meter may be shown in close-up, working synchronously
with the clicks. The effects microphone makes it possible to vary not
just the volume but also the quality of sound, so that the clicks could
be soft and unobtrusive when they are not central to the discussion,
and crisp and strong when they become the centre of interest.
As in radio, recorded effects are used to suggest atmosphere. With
crowd effects plus a camera track in over the heads of six extras, the
viewer may be convinced it was shot in a crowded bar. Film may
establish a wide street scene, then its sound may continue over a
studio shot showing a small part of the action within it.
Television introduces an extra dimension to sound effects that radio
has never had, the ability to jump easily between different viewpoints ± say, between interior and exterior of a car or train. For fastcut sequences it is convenient to use parallel effects tracks with
interior on one and exterior on the other. The operator simply fades
quickly between them on each cut or uses a device that automatically links sound and vision cuts. For an effect in stereo, spreaders or
stereo reverberation may be added, or a multitrack recording used. In
any case, it is useful to pre-record effects sequences for complex
scenes, for which little studio rehearsal time may be available.
Footsteps in vision
Where footsteps are to be heard in the studio, the television or film
designer should help sound by providing suitable surfaces. Stock
rostra are often constructed with a folding frame to support a surface
that gives a characteristic hollow wooden sound to footsteps. For an
action area a suitable treatment would be to `felt and clad': this adds
a layer of thick felt, with hardboard on top, so that footsteps are
audible but no longer with the objectionable drumming sound. If
little or no footsteps are to be heard, add a further soft layer of rug or
carpet.
For `outdoor' areas the use of a sufficiently large area of peat not
only deadens the surface but may also deaden the entire local
acoustic. But certain materials commonly found outside cannot be
260
Sound effects
used in television studios. Sand cannot be used, as it spreads too
easily, gets in technical equipment, and makes an unpleasant noise
when a thin layer is ground underfoot. Sawdust is a convenient
substitute. For similar technical reasons, salt and certain types of
artificial snow cannot be used: again substitutes that do not cause
sound problems should be found.
Rocks are rarely used in studios, as they are heavy and inconvenient.
Wooden frames covered with wire netting and surfaced with painted
hessian have been used, as have fibreglass mouldings or expanded
polystyrene carved into suitable shapes. Unfortunately, none of these
makes anything like a satisfactory sound, so any climbing, rubbing
or bumping must be done with the utmost care by the performers ±
and then preferably only at places that have been specially arranged
not to resonate, perhaps by using peat or sawdust either as a surface
layer or packed in sacks. Falling `rocks' of expanded polystyrene
must have recorded sound effects exactly synchronized ± which is
easier if the action is seen in big close-up, with the blocks falling
through frame. Expanded polystyrene can sometimes squeak when
trodden or sat on, and particularly when small pieces are ground
against a hard studio floor. In such cases, post-synchronization is
safer: filming or pre-recording and tracklaying appropriate sound
avoids it completely.
Sidewalk paving slabs are often simulated by fibreglass mouldings,
as these are light and convenient to hold in stock. When laid in the
studio they should stand on cloth or hessian to stop any rocking and
to damp the resonance. If, in addition, performers wear soft-soled
shoes and there is recorded atmosphere, the result should be satisfactory. If clear synchronous footsteps are required, there is no good
alternative to real paving stones, despite the inconvenience and weight.
Stone floors can be simulated by the use of a special 12 mm (1¤2 in)
simulated stone composition laid on underfelt. But `stone' steps that
are actually made of wood are likely to cause trouble if any sound
effect is needed (and even more if not). If a close effects microphone
with good separation can be used for the footsteps there is some
chance of improvement, as heavy bass-cut can then also be used.
A dramatized series about a prison in a castle had trouble throughout a
television season with its simulated cobbles and fibreglass doorsteps
around its courtyard. For a second run, the film set was constructed
using real cobbles and stone ± which, to everyone's surprise, turned
out better not only for sound, but also for the programme budget.
Wind, rain and fire in vision
The sound effects for wind, rain and fire create no problem: recordings are used as in radio. But when they are combined with visual
effects the device that makes the right picture sometimes makes the
wrong noise, which must therefore be suppressed as much as possible, to be replaced by the proper recorded effect.
Sound effects
261
Wind in vision is generated by machines varying in size from very large
propeller fans about 2.5 m (8 or 9 ft) in diameter, through medium
sizes, 1 m (3 or 4 ft) across, and which can be moved round on a trolley,
down to small machines at 45 cm (18 in), which can be set on a lamp
stand and used for localized effects. The bigger sizes are very noisy and
push out a large volume of air over much of the studio (if used indoors).
Windshields are needed for any microphones within range. The largest
machines may be used for the individual shots of a film but are
inconvenient in the television studio, where the smallest size is best.
When run at low speed, the noise from them can actually sound like
wind (at higher speeds blade hum is heard). The visual effect is shortrange and the airflow can be controlled to avoid microphones.
Rain, when produced in the television studio, may be far too noisy, so
it should be arranged for the water to fall on soft material in a tray with
good drainage. Also, if it is in the foreground of the shot the feed pipe
may be close to the ideal microphone position, making balance on a
boom difficult. Another problem is that it takes time to get going and
even longer to stop completely, to the last drip. This may be overcome
either by slow mixes to and from effects (including recorded effects)
accompanying the transition to and from the rain scene or, more
commonly, by discontinuous recording and subsequent editing.
Fire ± including coal and log fires ± in television studios is usually
provided by gas, with any visual flicker effects added by lighting.
Sound equipment in vision
When radios, television sets, tape or video recorders or record players
are used as props by actors in television programmes they are rarely
practical: obviously, the sound gallery wants to retain control of sound
quality (by filtering, if necessary) and to maintain a good mix at all
times, and also to ensure perfect cueing in and out. Careful rehearsal is
needed for exact synchronization of the action. For any sound cues to
the actors, it is fed at low level to a studio loudspeaker.
Pre-recordings using the same studio are sometimes used for out-ofvision sound. This may help to get it in the correct perspective or,
more often, is used just to fix one element in an otherwise all-tooflexible medium. Music is often pre-recorded for complicated song
and dance numbers ± but this may be in a different acoustic and with
a close balance, both of which create matching problems.
Television sports and events
Sports and events create sounds that must be captured on television,
preferably in stereo. There are two main types:
.
Arena events. For these, crowd atmosphere and reaction are taken
in stereo and spread over the full available width. Action noise,
262
Sound effects
.
such as bat on ball, is taken on more highly directional microphones and usually placed in the centre. Sometimes an extra
stereo pair is used for action, but this is rarely crucial.
Processional events, where cameras and microphones are set at a
sequence of points along a route. Microphones are usually also
set to follow the action, using stereo according to the same
conventions as for picture. For television, it is not always necessary for sound to follow action precisely, provided left and right
match in picture ± it is better to concentrate on dramatic effect.
A racing car approaching makes little sound, but its departure
is loud. In mono the tail of the previous sound was often allowed
to tail over the start of the next (approaching) shot, but in stereo
a cleaner cut, and near-silence (apart from stereo atmosphere)
on the approach, is effective.
For very precise full-frequency-range pick-up of effects at a distance, with highly effective noise rejection 15 or more off-axis,
consider using an adaptive array combination microphone with its
digitally processed forward lobe that is only 8 across (switchable to
8 30 ). This might also be useful for pinpointing small regions
with a large crowd or (closer) enhancing the audibility of distant
dialogue in noisy surroundings ± for both television and film.
Sound effects on film
On film, effects are added in post-production either to maintain a
continuous background of sound or to define a particular quality of
near silence. In addition, effects of the `spot' type are made easier by
the process of synthesis that is used to create films from many
individually directed short scenes, within each of which the sound
may be given greater care than in the longer run of the television
studio. A live action effect that is recorded synchronously as a scene
is filmed can be enhanced in a number of ways:
.
.
.
.
it may be changed in volume;
a copy of the effect may be laid alongside it on another track, so it
can be changed in quality and volume to reinforce the original
sound;
the timing may be altered ± e.g. when an explosion has been
filmed from a distance and the original recording is out of sync
because of the time it takes for sound to reach the microphone,
the noise can be resynchronized visually;
the original effect may be replaced or overlaid by a totally
different recording, again in exact synchronization with picture.
In the film industry, the provision of footsteps and other spot effects
may be a separate stage in post-production, and is the province of
spot-effects specialists, often called foley artists. Their skill permits
the film director to concentrate on picture and action in locations
where there is extraneous noise from traffic, aircraft, production
machinery or direction, or where acoustics or floors are inappropriate.
Sound effects
263
The tracks are recorded to picture and can be moved, treated and premixed as required, before the final re-recording.
Even if an action sound works well on the main microphone, it may
be even better if recorded again as wildtrack (i.e. without picture)
and given its own special balance. When neither works satisfactorily,
some other spot sound simulating the proper noise may be used.
For example, take the case of a bulldozer knocking down a street
sign. In the main sync recording the dominant noise is engine and
machine noise from the bulldozer: the impact of blade on metal is
hardly heard. So a second version of the noise of impact is simulated, perhaps by using a brick to strike the metal, and recorded
without any engine noise. This is laid on another track and controlled separately. If the bulldozer knocks down a wall the sound is
complicated even more, and the dramatic importance of three sounds
may be the inverse of their actual volumes: impact noise may again
be the most important; then the falling masonry; then the bulldozer
engine and mechanical noise. Several recordings of the principal
event could be made with different microphone positions (on either
side of the wall), perhaps with additional sounds recorded separately, to be incorporated later.
Commercials and documentary features sometimes make use
of punctuating sound effects. For example, an object built up stage
by stage by stop-action techniques has a series of parts suddenly
appearing in vision, and each time this happens there may be a sharp
noise ± a recorded `snap-in', a tapped metal surface, a swish or a click.
Cash-register noises accompanying messages about the amazingly
low price of a product could be too much: such music may delight
the advertiser more than the customer. But there is a vast potential for
non-representational, comic or cod effects in commercials.
Chapter 16
The virtual studio
The studio without a studio: welcome to the virtual studio! There is
no dedicated acoustic space, just one room where the electrical analogue of sound is created, assembled, manipulated (or processed),
mixed and recorded. Real sounds may still be captured by real
microphones, but are not at the heart of the operation: they might
be recorded during a moment of quiet in the same control space, or
they may be imported.
Today, a virtual studio is most likely to be centred on a computer
with software that duplicates many of the functions of a real studio ±
and in its simplest (or purest) form that is all there is. Even the
recording process stays within the computer, although if the endproduct is to be, say, a CD, the computer turns into a digital audio
workstation, and the CD is `burned' (a term that recalls the record's
hot origin under a laser beam) within the same computer. Given the
vast and seemingly ever-increasing processing power and memory
plus built-in hardware provided inexpensively even in home
computers, this is commonplace. The software, including programs
on cards that need to be slotted in to the back of the computer,
certainly adds to the cost, but the total will still be far less than much
of the hardware needed in a `real' studio.
Virtual studios and their kin, project studios, did not start out like
this. Their (for many current users) misty origins lie in an era when
digital technology was little used. The electrical analogues of sound
remained robustly analogue, and the only digits employed were
the fingers that turned knobs and operated the switches that literally and laboriously manipulated the processes. One of several early
pathways towards the virtual studio was followed mainly by
traditionally-skilled European composers who wanted to explore
the musical possibilities offered by new technology in order to
pioneer their `modern' music. In Britain the same approach was
used to create a combination of music and effects (for radio and
television) that was called `radiophonics'. Since these gave rise to
The virtual studio
265
many of the techniques and operations in use today, along with
others now less used, but which might still suggest new possibilities,
some of this history is described in Chapter 17.
A second and ultimately more powerful influence on the evolution
of the virtual studio came directly from what digital technology
could offer. Crucial developments along this pathway included the
sequencer, which could control a synthesizer or drum machine, and
early samplers (initially with very limited capacity) from which the
output could be fed to a multitrack recorder, at that time on analogue
tape. Even at the stage when such devices were little more than toys
in makeshift home studios, it was clear that little more than these
basics was needed to make some kind of music. Originally, all of
the components with their individual processors were in separate
boxes that needed to be connected in some way. This required the
development of a commonly accepted `musical instrument digital
interface' ± MIDI.
The key developments along this more popular pathway to the
virtual studio were nearly all direct consequences of increasing
processing power and memory, together with the inventiveness of
the programmers using it. Sequencing programs were written for
home computers that could trigger sounds from the sampler and
synchronizer and sync them to a drum machine's (separately) programmed rhythms. The samplers might also have built-in programs,
perhaps to change pitch or duration, and a keyboard could be
persuaded to play the often rather irritating results. Effects, too,
could be sampled and laid down on spare tracks on a multitrack
recorder, prior to mix-down. Dance music produced by such minimal
means became immensely popular, making celebrities of, and
fortunes for, its most imaginative creators and users.
[16.1] Sequencer timeline screen
The central workspace in a virtual
studio is a screen that shows audio
and MIDI tracks below a timeline that
is likely to be of `step time' (based on
the length of musical notes) rather
than clock time (from time code). In
this example, audio and MIDI are
similar in appearance, except that the
audio (an actual recording) can be
represented by its waveform, while
the MIDI parts (which contain only
control data) are shown as
rectangular blocks. A smaller
transport window, with controls like
those of a digital tape recorder, would
be inset. The example shown here
and below is Logic ± one of a small
number of software packages that
are still described as sequencers, but
which have developed into powerful
music composition and audio
production systems.
In due course, all the separate boxes were gobbled up by the
computer, although their functions still exist within its programs.
The computer itself became the sequencer, sampler, synthesizer,
drum machine and rhythm generator, plus multitrack recorder ±
and much else. Digital editing emerged almost serendipitously as
an extension of sampling, as samples got bigger and the programs to
manipulate them grew ever more complex. Almost by the way, these
same developments engulfed and swallowed up the previously
separate fields of modern classical composition and synthesized music
for radio, television and film.
The next major addition to the virtual studio, as to regular studios,
has been digital signal processing (DSP). One form in which this
may appear is on a `platform' within the computer. A manufacturer
may offer `a total production environment' with digital recording,
sequencing, editing, mixing, mastering and a variety of signal
processors, plus the platform for `plug-ins' by other software developers. This is much like the way in which studio hardware is centred
on a particular console, to which racks full of specialized equipment
are literally plugged in.
The visual display screen that comes with a plug-in often represents
a set of rack-face controls that an operator might recognize and
266
The virtual studio
know how to use. But use them how? It may not be difficult to slide
faders and turn knobs with a mouse, but is this the best way to do it?
From a state where everything is hidden within the computer with
simulated controls on-screen, the seemingly natural next step is a
return to at least some of the physical equivalents, using control
surface hardware that is reattached to the computer.
Take this far enough and the ultimate reattachment might include a
more sophisticated control surface (we might call this a `console'),
a few microphones, and for the sound sources, an acousticallydedicated space. In reverse, this is just what happens when a regular
studio has a MIDI area within it. In a drama studio, it is convenient
to build up complex effects sequences with their timing and pitch
played in from a MIDI keyboard, then processed and pre-mixed on
its associated control surface. The same keyboard can also provide
musical effects.
The more common type of virtual studio is the musical equivalent of
a home office. At its heart is a good computer and screen, and often
an electronic musical keyboard. One other essential is a good loudspeaker system. But given that much of the background music heard
on radio and television, and some on DVD or film, is constructed in
this way, the more ambitious studio also has 5.1 surround sound.
(One detail in the layout for this is that in such a confined space the
subwoofer needs to be positioned with some care. One suggestion on
how to find the best spot is: check where in the room the bass is
weakest and put it there.)
Readers who want to set up and use a virtual studio can find plenty
of advice in the books, magazines and websites devoted to this field
and specialized aspects of it. But for those concerned with the
operational techniques employed in the sound studios that are the
primary subject of this book, this chapter offers a brief introduction
to components of the virtual studio that are increasingly used in all
studios.
MIDI plus audio
[16.2] Opening audio or MIDI parts
Clicking on a segment of an audio
track opens a window that shows
more detail: zoom in to see the
waveform in close-up. Similarly,
clicking a MIDI part may reveal
the keyboard of an instrument and
the notes it is instructed to produce.
(Other display options include lists
or musical notation.)
Originally coined as the name for a way of linking a few things
together, `MIDI' is now understood by many of its users to encompass far more than that ± in fact, to include many of the linked
objects plus all that is done with them. Tracks associated with digital
audio and MIDI events can be shown one above the other on the
same screen. They work together and can be played together (subject
to a timing correction to adjust for a slight delay in MIDI play-out).
`MIDI ‡ audio' has become a field in its own right, with the usual
abundance of literature to consult.
The signature component of a MIDI system is the keyboard derived
directly from the synthesizer, and indirectly from the piano, the
black and white key layout of which is copied, but extending generally
over a more modest five octaves. The processors and controls that go
The virtual studio
267
with this may fit either in the same box (as in many domestic MIDI
systems) or on a separate mini-console. A virtual studio no longer
needs a physical keyboard and controls ± these can be mimicked
on the screen of the computer ± but they are more convenient if
used often. Also, many users do feel more comfortable, or better
rewarded, by having the hands-on keys, faders and switches as
well as the screen displays.
With MIDI, keyboards can be used either in a conventional mode
with keys used for controlling pitch, or with a series of samples
triggered by the keys, or with each key designated separately for
some other control function. It follows that either pitch-perfect
conventional scales or unconventional progressions can also be
constructed and passed through it. An external MIDI keyboard,
sequencer, drum machine or synthesizer can be linked to a computer
by MIDI cables and a MIDI interface (the MIDI equivalent of a
sound card) that translates the stream of MIDI data into a form the
computer can work with.
[16.3] Track mixer window
Opening a mixer window over the
timeline window reveals faders
spread out like a mixer console across
the screen, with other controls in line
above the faders. Each fader is
numbered (and may also be named)
to match the list of tracks, and each
fader operates directly upon its
corresponding track. The MIDI strips
are simpler than those for audio,
which must include EQ, `sends' to
various busses, and insert points for
AR, compressors and limiters, and
more specialized DSP plug-ins that
may be assigned to them. Audio can
be either mono or stereo, perhaps
with a meter for each leg next to the
fader, and it may be possible to mix
to 5.1 surround.
In its early incarnation as a synthesizer, the MIDI keyboard was
closely related to the electric organ family, in which musical notes
with various qualities were created, with pitch and volume controlled conventionally by the keys. In the software version of this,
many of the synthesized notes are replaced by samples created by
skilled musicians on real instruments taken either from a sampler
program or recorded locally ± so the keyboard is used not only for
notes from the regular keyboard instruments, but also for strings,
woodwind and brass. Full drum-kits can be included. Other programs lay down rhythm, allow a beat or drum taps to be replaced by
musical tones or sources, snap off-key voices to perfect pitch and
enrich or add depth to their harmonic structure.
Put a chord or random notes in an `arpeggiator' and it will arrange
for them to ascend or descend with metrical precision. Or `quantize'
the notes (or drum taps) to set them in some perfect but expressionfree tempo. One way to improve on that is to render it graphically as
a full score, on which note order, length, position and musical
qualities can be changed at will. Mixing, at best, is fully automated.
Editing MIDI ‡ audio is similar to that described in Chapter 18.
MIDI ‡ audio sequencers are now immensely (and ever-increasingly)
powerful and versatile packages. The main names are Logic Audio,
Digital Performer and Cubase VST (for `Virtual Studio Technology').
Plug-ins
A `plug-in' may be completely metaphorical, consisting of a disk
that contains the program: this is loaded rather than plugged in to
anything, and then runs on the host computer. Or it may be real, a
card with a dedicated chip that carries the program: this is literally
plugged in to a slot in the computer and runs in parallel with it.
There are variations on these themes, but those are the main options.
268
The virtual studio
The computer, in turn, must be ready to receive the plug-in, with a
platform for it to link into, and through which it operates. Pro Tools
and Logic offer platforms into which over a hundred devices by
other manufacturers (as well as their own) can be plugged. Many
processes are already available with the platform, so they and thirdparty designers compete to offer some unique capability:
.
.
.
.
.
[16.4] DSP plug-ins
The sequencer (production system)
may act as host to plug-ins, some
native (i.e. by the same
manufacturer), others from a wide
range of sources that use
a compatible format, such as
DigiDesign's TDM. The function of
a plug-in is usually characterized, at
least in part, by the graphic style that
appears in its window. A `modern'
style emphasizes a function that
requires (and has only become
possible with) substantial processing
power. Some reveal their musical
credentials with MIDI-style
keyboards.
[16.5] Synthesizers and samplers
Software-based instruments may be
directly available as sources alongside
the audio and MIDI input.
to make it possible to assemble any musical arrangement, of any
musical quality, with inhuman precision of metre or pitch, then
add musical expression and imperfections of timing by hand;
to emulate the characteristics of existing respected hardware
together with all its useful add-ons, plus any digital improvements;
to emulate historically successful (and often wildly expensive)
analogue devices that were once prized by studios and used to
create the distinctive sounds associated with famous performers;
to perform digital functions that were impracticable in analogue;
to complete all the complex production processes that are necessary in, say, mastering media in a wide range of digital formats.
Some are narrowly defined, for some specialized operation; others
come in generous DSP bundles, offering the quality or value that
users associate with a particular name. The categories (or functions)
include the following:
AR. For some of the earliest DSP hardware, specialist designers
created distinctive `reverb' packages, of which these plug-ins are
just the latest versions. The only difference should be in how
the controls are presented. One kit, to be different, calls itself `spatial
effects'.
EQ. Filters were some of the earliest analogue devices, with the
bonus that the simpler they were, the smoother their operation: a
circuit containing just one resistor and one capacitor curved
smoothly into a roll-off of 6 dB per octave. Most EQ plug-ins
emphasize their `analogue quality of sound', meaning gentle,
smooth transitions unless otherwise stated. Some emulate particular
early rack-mounted EQ kits, with photographic screen controls to
match, one specifying the simulated use of tube (valve) technology:
`engineers run tracks through this even without EQ, to make them
sound better'. Some offer `sweep' effects. Parametric equalizers
allow the frequency, bandwidth and gain or loss to be varied
continuously, not just at specified settings. Some add presence or
enhance bass `without increase in level' and also throw in de-essing
(reduction of vocal sibilance due to HF peaks).
Compressors, expanders, limiters and gates. Some emulate `classic'
analogue equipment, again with photographically accurate screens.
Some even simulate the way in which analogue overload introduced
distinctive forms of distortion, so that digital mixes do not
completely lose this characteristic ingredient. Others offer digital
perfection to fend off the potential disaster of digital clipping. To lift
the volume of a track as high as possible, it is necessary to check for
ephemeral but potentially damaging peaks, delay the signal, then
stop them happening.
The virtual studio
269
Synthesizers. As ever, these are likely to include oscillators, filters,
envelope shapers, chords, and a wide variety of delay effects such as
chorusing, flanging, phasing, multi-tap and `ping-pong' repetitions;
also `science fiction' effects on speech such as ring modulator and
Vocoder (the analogue origins and uses of these are described in
Chapter 17). Some plug-ins offer rotating loudspeaker effects or
emulate vintage favourites such as the Fairchild or Moog (rhymes
with vogue). `Frame accuracy' may be specified ± i.e. no apparently
random tiny delays.
Spectral delay. This splits each audio input channel into 1024
separate frequency bands so that delay time and other parameters
can be adjusted independently on up to 160 of them at any one time.
[16.6] `Retro' plug-ins
Many DSP plug-ins glory in their
`retro' origins by graphic or even
photographic displays based on the
control panels of archaic but
treasured analogue devices ± in some
cases even to the extent of showing a
simulated tape loop or trailing patchbay connectors. Despite the period
style that recalls the limited capacity
of the originals, the digital programs
often require prodigious processing
power.
[16.7] Reverberation insert
When AR is assigned to an insert tab
above a fader, clicking it brings up a
window showing controls for the
selected program. This one is native
to the system shown above, and like
most operations within it, can be
automated. Other manufacturers'
AR software ± and many other
plug-ins ± can also be mounted (using,
e.g., TDM, DigiDesign's Time Division
Multiplex format) and operated, often
also with full automation, from the
same group of buttons.
Retro quality effects. A number of packages glory in `down-processing' or the creation of `lo-fi' effects. These include old guitar
amplifier and cabinet effects that are all too easily eliminated by
digital technology, together with the deliberate distortion of `buzz',
etc., often added by the player; also variable degrees of tube (valve)
`warmth' to modify drums, vocals, synthesizers, etc. Some add preamplifier overload or tape saturation ± or the more benign quality of
`vinyl groove'.
Noise reduction. A field in which digital technology excels. This
reduces mains hum and its harmonics, air-conditioning, computer
whistles, etc. It also restores archive recordings, removing (most of)
the clicks, hiss or rumble. These repairs all need to be applied with
care: find a setting at which side-effects appear and set back from it.
Pitch and time. Another winner for digital. A number of packages
offer variations on pitch correction. Some are designed to `fix'
voices or instruments that are off-key. Also available are pitch
transposition and delay. Vocal formants may be shifted, changing
apparent gender or even species. Time changes, with or without
change of pitch, are usually accompanied by a claim of `no chipmunk effect', which means that the repeated tiny overlaps are
effectively disguised. A related set of effects enriches the timbre of
voices by adding harmonics.
Microphone modelling. This converts the frequency characteristics
of one named microphone to another. It can also take into account
polar response (change of response with angle), distance (for proximity effects) and losses due to windshields. It is based on manufacturers' data on both modern and classic types (i.e. with tube
preamplifiers).
Sampling. This is often supplemented by the provision of sample
libraries, which may include a wide range of either Western or
ethnic musical instruments, or specialize in a particular department
such as drum-kits, percussion, saxophones or a range of `classic'
electric guitars. Each pitched instrument is sampled at a succession
of points on the musical scale. Because intervening notes are created
(generally) by playing them at different speeds, the more original
samples there are the better. Another variable is playing styles: some
offer sounds created by named players that are copyright-free.
270
The virtual studio
(Taking samples from other sources risks copyright infringement.)
Effects libraries are also available in this form. MIDI keyboards may
be used with all these.
Also available as plug-ins are technical processes such as word
length reduction, dithering and mastering tools. Typically, these
encode 24-bit audio for 16-bit media such as CDs. There are
surround sound tools and surround metering (for stereo, LCR,
LCRS, quad, 5.0, 5.1, 6.0, 6.1, 7.0 and 7.1) and a program for
downmixing 5.1 to headphones without killing it. Also available
are phase monitoring and alignment tools (to resynchronize or finely
control the timing of tracks) and electrical measurement displays.
Control surfaces
[16.8] Musical notation
Traditional notation (available as part
of sequencer programs) may be used
either to convert a MIDI output
display into sheet music or to use it
directly as a composition editing tool.
Adding a small manual mixer, perhaps with eight motorized faders
and a master, a selection of function keys, buttons (to select, mute,
record or solo on each channel), plus a jog and shuttle (scrub) wheel
to wind back and forth against the `playhead' marker, makes the
operation of a virtual studio smoother, faster and more intuitively
`hands on'. Rotary potentiometers, with displays to show selected
parameter names and values, may be used to control and automate
panning, EQ, dynamics, send levels and other functions on DSP
plug-ins. All this may cost little more than some of the software, though
each added eight-channel extension costs almost as much again.
For those who want a larger control surface that includes most (or
all) of the functions of a multi-layered fully-digital console,
complete with automated faders and an extended range of digital
processing, this might cost about as much as a well-appointed car.
This makes it rather expensive for the hobbyist, about right for much
project-studio work, and cheap compared with many broadcast or
recording studio consoles. In a virtual studio, where tracks are
usually laid down sequentially, the fussiness of layering is less of a
disadvantage than it would be in a regular studio, where it is more
convenient to be able to reach any channel instantly, without switching layers. And it takes up less space.
Digital problems
[16.9] Hardware controller
A compact unit that features full
transport controls, automated
touch-sensitive faders (eight channels
and a master), rotary potentiometers
and displays to control selected
(external) DSP functions, and a scrub
wheel. This can be extended by
adding further eight-channel fader
modules.
One of the most important questions about any digital platform is:
how stable is it? The next is: what do you do when it falls over?
These questions are not limited to the virtual studio, but it is as a
good a place as any to ask them.
Computers crash. This is a simple fact of life for anyone who has
dealings with them. Computers that are asked to do complicated new
things (or sometimes even quite simple things that they were not
doing last week) crash in ways and at times that even the most
The virtual studio
271
tolerant can find frustrating. Computers used in broadcasting or in
other time-limited projects require a history of stability ± that is, it
must be known from experience that they do not crash very often.
They also require redundancy, together with automatic, instant or at
least rapid switching so that problems can be bypassed and sorted
out later. Project work is (sometimes) more tolerant of such problems
than broadcasting is. But tolerance diminishes in the presence of
paying clients. These may be vital considerations when setting up
any computer-driven system.
Another common problem associated with computers is psychological: the danger of getting locked into a search for bugs (or
simply for the correct instructions) that to any observer can seem
incredibly time-consuming. It is, of course, essential to be sufficiently familiar with programming that no time is wasted, and to
ensure that a crashed system can be reinstated quickly. Consoles
with computers should automatically store information at frequent
intervals against this possibility, but even so, the recovery procedure
requires enough rehearsal to ensure that it does not significantly
reduce production time. Handled (and operating) well, computerdriven digital audio is a dream come true; handled badly or failing
for perversely obscure reasons, it is a nightmare that makes the
apparently simple, understandable faults of analogue audio seem
almost forgivable.
Whether in the virtual studio or not, increased digital processing
brings with it more and more `black boxes', items of equipment that
most operators cannot repair in the event of failure. Inevitably,
operators become the dedicated users of proprietary systems. And
new developments in these systems (which appear all the time)
require new software, which can all too often lead to more bugs or
unfamiliar modes of breakdown. The best time for any user to move
in is when a generation of equipment has become reasonably well
established, rather than at the cutting edge of the next great advance ±
unless that advance is vital to the effects required, and the user is
happy to help refine the system. Intermittent users of such technology
are liable to find it difficult to keep up with what is offered.
In digital technology, rapid turnover is another fact of life. So although
individual items are often much cheaper than their analogue forebears,
the costs even out. But just occasionally redundant hardware may
find a second use. One software item among all those plug-ins
converts an outdated G3 Mac computer into a `high performance
standalone non-linear virtual VTR' for use in a digital video production environment.
In general, the shift to digital technology has made possible a
greater emphasis on more analytical techniques, towards perfection
rather than performance, so project work is likely to be more
successful and rewarding when this is the main objective. Fortunately
for the users of analogue equipment, the technology time-lag is
longer. A microphone, for example, can have a long useful life,
and fashions in its use can even bring apparently dead equipment
back to life. Or ± if it really is dead ± fashion may dictate its digital
reincarnation.
Chapter 17
Shaping sound
This chapter offers a short digression, to revisit some predecessors of
today's virtual studio. The technology was far less powerful, but this
did not limit the imagination of its users. Some of what was achieved
by slow and complicated analogue procedures is now available as
just a small selection from the huge range of facilities on digital
signal processors (DSPs), and the names for the effects (often
reflecting their analogue origins) are unchanged.
Here we will take a look at how developments in audio technology
opened up new ways of shaping sound, and review a few that,
although now in decline, might still have uses or could possibly lead
in new directions. Long before the digital era, it was clear that
anyone with a microphone, recorder and a razor blade (to physically
cut and edit tape) could show that the formal organization of any
sequence of electronic or natural sounds has some resemblance to
music, although how good that was depended on the musical sense
of the person making the arrangement.
Some of the methods by which music is now synthesized in the
virtual studio were first developed in the earlier field of electronic
music (at its purest and most austere, in Germany). Similarly,
sampled sound began life as musique concreÁte (as the name suggests,
in France). These were both used mainly by composers who wanted
to experiment with new musical forms. In Britain the BBC, seeking
a more practical use for these exotic ways of shaping sound, applied
them to the creation of radiophonic music and effects for radio and
television. Some of the ideas that underlie these (old) new departures
are outlined below.
Practically every sound we hear reproduced ± however high the
quality ± is a rearrangement of reality according to particular
conventions. For example, monophonic sound is distorted sound in
every case except when the effective point source of the loudspeaker
reproduces a sound that was originally confined to a similarly small
Shaping sound
[17.1] Classification of sound
The dynamics or envelopes of
sounds: early classifications devised
by musique concreÁte composers and
engineers as a first step towards a
system of notation. The terms shown
here could represent either the
quality of an original sound or the way
it might be transformed.
Attack. 1, Shock: bump, clink, etc.
2, Percussive: tapping glasses, etc.
3, Explosive: giving a blasting effect.
4, Progressive: a gradual rise. 5, Level:
at full intensity from the start.
Internal dynamics. 1, Steady: an
even quality. 2, Vibratory or puffing.
3, Scraping or tremulous. 4, Pulsing:
overlapping repetitions blending
together. 5, Clustered subjects
placed end to end.
273
source. Every balance, every carefully contrived mixture of direct
and indirect sound, is a deliberate distortion of what we would hear
in the same sound, live. This is accepted so easily as the conventional form of the medium that few people are even aware of it.
Perhaps distortion would be better defined as `unwanted changes of
quality', rather as noise is sometimes defined as `unwanted sound'.
Sound is continuously being `shaped', and the listener is always
required to accept some convention for its use.
When an audience is presented with some `new sound', good listening
conditions are even more important than for conventional music.
A listener with good musical imagination who hears known instruments
and established musical forms can judge the quality of composition
or performance in spite of bad recording or reproduction. Although
they reduce enjoyment, noise and distortion are disregarded. The
musical elements are identified by reference to the listener's knowledge and experience: imagination reconstitutes the original musical
sound.
Plainly, this does not happen on the first few hearings of music in
which the new sound is produced by what, in other situations, might
be regarded as forms of distortion. For this, imperfections produced
by the equipment are, for the audience, indistinguishable from the
process of musical composition: they become part of the music.
A major difference between synthesized or sampled recordings and
the conventional music that audiences are more familiar with is to
do with performance. The traditional composer commands a large
number of independent variables: the choice of each instrument and
the way it is to be played; the pitch of each note and its duration; the
relationship in time and pitch between successive notes and between
different instruments. But however much is specified, there is always
room for interpretation by performers. In contrast, an electronic or
concrete composition exists only in its final state, as a painting or
carving does. Unless it is combined with conventionally performed
music or created in real time, for example from a MIDI keyboard, it
is not performed: it is simply replayed. In this respect musique
concreÁte is like electronic music, from which it otherwise has certain
fundamental differences.
Musique concreÁte was made by sampling the existing timbres of a
variety of sound sources, transforming them, then assembling them
as a montage. Electronic or synthesized music began, as might be
expected, with electronically generated signals such as pure tones,
harmonic series or coloured noise (noise contained within some
frequency curve). The results are distinctively different.
Decay. 1, Reversed shock.
2, Accented. 3, Deadened.
4, Progressive. 5, Flat: cut off sharply.
One characteristic quality of sampled sound is that the pitch relationship between the fundamental and upper partials of each constituent
remains unaltered by simple transformation unless any frequency
band is filtered out completely. This is both a strength and a restriction: it helps to give unity to a work derived from a single sound or
range of sounds, but severely hampers development.
In electronic music, on the other hand, where every sound is created
individually, the harmonic structure of each new element of a work
274
Shaping sound
[17.2] Line and band spectra
Instantaneous content: this sound is
composed of individual tones
superimposed on a broad-spectrum
band of noise.
is, in principle, completely free. Where musique concreÁte has
restrictions, electronic music offers freedom ± an essential difference
between the two. But together, these techniques and their successors
have a proven value in extending the range of sound comment. This
developed as a field in which sound effects took on formal musical
qualities in structure and arrangement, and musical elements
(whether from conventional or electronic sources) complemented
them by taking on the qualities of sound effects. A useful attribute
for anyone working in this field is a strongly self-critical sense of
humour.
Musique concreÁte
Musique concreÁte, according to its originators, is (or was) built up in
three distinct phases: selection, treatment and montage.
The first characteristic of musique concreÁte lies in its raw material:
a sampled sound which, by the choice of beginning and end, becomes
complete in itself. In some of the earliest compositions, emotional
associations of the original sound were incorporated into the music.
There was a later reaction away from this, on the grounds that the
lack of an immediate mental association could lend power to sounds.
The basic materials might include tin cans, fragments of speech, a
cough, canal boats chugging or snatches of Tibetan chant (all these
are in a work called Etude PatheÂtique). Musical instruments are not
taboo: one piece used a flute that was both played and struck.
Differences in balance or performance can also be used to extend
the range of materials. All this is very similar to the way that
samples integrated into popular music have included news actuality,
political statements and fragments of other people's compositions.
The preliminary concreÁte recording was described analytically
(again, by its originators) in terms of a variety of sound qualities:
.
.
.
instantaneous content ± frequency spectrum or timbre (which might
contain separate harmonics, bands of noise or a mixture of the two);
the melodic sequence of successive sound structures;
its dynamics or envelope (the way sound intensity varies in time).
The second stage in building musique concreÁte was treatment of the
component materials to provide a series of what its composers called
`sound subjects', the bricks from which the work would be constructed. To help classify their growing range of `manipulations', the
Groupe de Recherche de Musique ConcreÁte distinguished between:
.
[17.3] Musique concreÁte:
transformation of dynamics
(Original sound in centre.)
.
Transmutation of the structure of sound ± changing instantaneous
sound content to affect melodic and harmonic qualities but not
dynamics. This includes transpositions of pitch (whether continuous
or not) and filtering (to vary harmonic structure or coloration).
Discontinuous transformation of the constituent parts of a sound
by editing ± a sound element could be dissected into attack, body
and decay (today, divided further into attack, decay, sustain,
Shaping sound
.
275
release ± `ADSR'), then subjected to reversal, contraction, permutation or substitution. This form of manipulation varies the dynamics
of the sound, though at any instant the content is not altered.
Continuous transformation, in which the envelope of a sound is
varied without editing, by the use of faders, AR, etc.
In the third phase, the construction of musique concreÁte, the sound
subjects were put together piece by piece, like the shots of a film.
One method was to edit a single tape by cutting and joining pieces
together (montage); another was dubbing and mixing in the way that
sound effects sequences might be assembled. Today, they would
simply lay tracks in digital memory and mix them down.
In the early days of musique concreÁte the only way to change the
pitch of a sound was to change the speed of replay on tape or disc.
This led to a more versatile analogue pitch changer in which the
signal was recorded on the surface of a rotating drum: with this,
pitch and duration could be changed independently ± but with a
characteristic `chipmunk' effect from the overlapping segments.
Happily this monstrous device has been superseded by a wide range
of digital signal processors (see Chapters 12 and 16).
Another analogue device that could change pitch independently of
duration was the ring modulator. This allows a fixed number of
cycles per second to be added to, or subtracted from, all frequencies
present (e.g. components f1, f2, f3 become x ‡ f1, x ‡ f2, x ‡ f3,
whereas simple alteration of speed multiplies all frequencies by a
common factor). Frequencies could also be inverted (x f1, x f2,
and so on) so that high becomes low and vice versa. With this type of
`transmutation' exotic qualities of consonance and dissonance are
obtained. They do not occur naturally (in conventional music) but
have been used in some electronic music. Again, this effect is now
available on digital signal processors.
Modulators have been used in telephony, so that narrow-frequency
bands containing the main characteristics of speech can be stacked
one above the other on a broadband line. In their modulated form
some speech bands may be transmitted in an inverted form ± and it is
possible to bring them back down to normal frequencies while still
inverted. For example, a band of speech might be inverted so that
what was 2500 Hz becomes zero, and vice versa. The most striking
result of this is to invert the three main bands of formants, so the
speech is translated into a sort of upside-down jabberwocky
language ± which, curiously, it is actually possible to learn.
Among more complex effects, a filtered human voice can be used to
modulate a train siren to make it talk. The effect of an exaggerated
machine voice can be obtained by modulating an actor's deliberately
mechanical-sounding speech with a tone of 10±20 Hz. In electronic
music, two relatively simple sound structures can be combined to
create a far more complex form. After using a modulator, it may be
necessary to use a filter to get rid of unwanted products.
[17.4] Musique concreÁte:
transmutation of frequency
(Original sound in centre.)
The Vocoder was designed to encode the basic parameters of speech,
including its tonal and fricative qualities, so that they could be
276
Shaping sound
imposed on any electronic or other source and give it the characteristics of a voice, including intelligible speech and song. This extended
the range of techniques available to films and television programmes
that specialize in alien voices and mechanical monsters, which had
previously depended heavily on ring modulators and swept filters
(the latter usually wobbles a filter up and down in frequency). The
Vocoder also offers possibilities for humanizing electronic music by
encoding some characteristic `imperfections' of musical instruments
and operating with them on electronic (or other) source material. It
could be used in conjunction with changes of pitch or duration, or
any other available effect.
When there is picture as well as sound, it is possible to link the two
directly, deriving or modulating an audio signal from some parameter
taken from the video signal; or alternatively, allowing audio to modulate
video or to control the generation or selection of picture elements.
Electronic sources
The creator of electronic music has, in principle, complete control of
resources, synthesizing every detail of the sound. With no performer
in the loop and therefore full responsibility for the finished work, the
composer has total freedom ± subject only to the limitations of time,
patience and equipment. In the early days that patience was needed.
The equipment for this includes sine tone oscillators for producing
single frequencies or combinations of tones, plus sawtooth and
square-wave generators for richer harmonic structures. The sawshaped wave contains a full harmonic series, while the square shape
includes only odd harmonics, like those of a clarinet. These could be
treated by `subtractive synthesis', i.e. filtering ± which was less
time-consuming than an additive process. Precise calibration was
essential: if musical instruments are off-pitch, only one performance
is affected and the score remains unaltered, but in electronic music
the first `realization' is likely to be the only evidence of the
composer's intentions.
[17.5] Transposition in pitch
When a sound is replayed at twice
the original speed it changes both the
timbre (A and B) and the dynamics
(C and D).
White noise is another type of sound used in electronic music and its
successors. It could be produced by amplifying the output from a
noisy vacuum tube: special tubes were made for the purpose. White
noise alone is not particularly useful, but again in combination with
filters it could be used to produce coloured noise, either of indeterminate pitch or related to a musical scale by placing peaks at
particular frequencies and troughs at others, or by using bands of
chosen widths. Another effect in early electronic music was
achieved by imposing a sinusoidal envelope (or any other waveform) on the noise to create anything from a series of puffs of sound
to a tone that could be used musically ± then slowed down again to
reveal its granular structure.
A characteristic sound in early electronic works was short bursts of
tone at various frequencies, sounding like a series of pips, beeps and
Shaping sound
277
bloops. If these were assembled from short lengths of tape, variations
in attack and decay could be introduced simply by changing the angle
at which the tape was cut. A 90 cut produced a click at the beginning
or end; other angles softened it. Other kinds of attack and decay
were created electronically, producing effects that were described by
relating them to those of a musical instrument of similar attack and
tone quality.
In parallel with the development of `pure' electronic music, there
had been developments of playable devices ± electronic organs ± with
names like Monochord, Melochord and Spinetta, besides the more
conventional Compton and Hammond organs. The Trautonium was
one of a group of instruments that could be used (and played live)
to provide notes of any pitch, not just those corresponding to a
keyboard. Organs, as one class, shade gradually into another, the
synthesizers. Most of the electronics of early synthesizers were
`analogue' also in the special sense that they copied or reconstructed
the required waveforms or noises by means that provided a continuous shape, which could also be continuously modified from the
start. The next stage of development was to add some digital
circuitry ± for example, by using a digital control device to program
musical sequences of treated sounds, so creating the first sequencers.
In its time the Moog was a revolutionary design of analogue synthesizer employing compact multifunction voltage-controlled oscillators
and amplifiers with what was seen as a generous ability to modify sound
structures in interesting ways. (Now it is found as a digital effect.)
The application of techniques derived from both electronic music
and musique concreÁte to broadcasting was pioneered in Britain. At
the BBC the product of its Radiophonic Workshop was applied
directly to the needs of general radio and television programming.
Since it began, in 1958, the BBC's Radiophonic Workshop has
updated its equipment many times, in a process of evolution that
has tracked the revolution in audio processing and recording until,
by the late 1980s, digital and MIDI techniques dominated the field.
This expertise has since dispersed, and around the world there are
now vast numbers of such studios. Very few are devoted to `pure'
experimental music, a minority taste. But along the way it has been
shown that completely new musical forms are possible. One example
is outlined below.
New laws for music
Most synthesized music is based on conventional musical scales:
these are directly related to human powers of perception and appreciation, which in turn are attuned to harmonic structures that arise
naturally. But by starting from pure electronic sources these musical
`laws of nature' can be relaxed ± or transgressed. Given how human
minds work, whether this is much use depends on the quality of
individual offerings.
278
Shaping sound
[17.6] Electronic music
Score of the type devised by
Stockhausen to show the structure of
his `Study II'. 1, Pitch: this shows the
groups of tone mixtures used. Each
block represents a group of five
tones. 2, Intervals of tape measured
in cm. The tape speed is 76 cm/s
(30 in/s). 3, Volume: dynamics of the
tone mixture.
[17.7] Tone mixtures
Five typical mixtures used in
Stockhausen's `Study II': 193 were
used in all.
All scales used in conventional music have one thing in common:
the interval of an octave, in which the frequency of the upper note is
exactly twice that of the lower. If the two are sounded together on
orchestral instruments the combination sounds pleasant because of
the concord between the fundamental of the higher note and the first
harmonic of the lower, and also between many other harmonics they
have in common. But if one of the notes is shifted off-pitch, these
concords are lost; they are present only when the frequency of the
two fundamentals can be expressed as a ratio of small whole numbers. Note also that without the harmonics there would be no sense
of concord or discord at all, unless the fundamentals were fairly
close together. If we start with two pure tones at the same frequency
and then increase the pitch of one of them, the first effect we hear is
a beat between them. As the difference becomes greater than 15 Hz
(two or three times this at high frequencies), the beat is replaced by
the sensation we call dissonance. This increases to a maximum and
then falls away, until as the frequency ratio approaches 5:6 it is no
longer apparent. For pure tones, dissonance occurs only within this
narrow range. (There is one proviso: that the tones are not loud. If
they are, the ear generates its own harmonics.)
Some experimenters with electronic music, noting all this, have
concluded that our conventional concept of scale is just a convenient
special case, and that if only we could create harmonic structures
that were not based on the octave, completely new scales could be
devised.
[17.8] Electronic music and
reverberation
Spectrogram of part of
Stockhausen's `Study II', showing the
softening effect of reverberation.
Each of the original groups was
replayed through a reverberation
chamber (digital AR was not yet
available).
An early example of electronic music based on a totally new concept
of scale has (exceptionally) been published as a complete score,
showing everything that had gone into the making of Karlheinz
Stockhausen's `Study II'. When the work is played to a listener with
perfect pitch, or even good relative pitch, it seems immediately that
`something is wrong' ± or at least, that something is different. By all
conventional musical standards, something is: its arrangement of
musical intervals did not exist before the introduction of electronic
sound synthesis.
Shaping sound
279
A conventional scale accommodates as many small-whole-number
relationships as possible by dividing the octave into 12 equal, or
roughly equal, parts. Each note in the chromatic scale is about 6%
higher in frequency than the one before, so that the product of a
dozen of these ratios is two: the octave. But Stockhausen, for his
`Study II', devised instead a completely new scale based on an
interval of two octaves and a third. This is the interval between a
note and its `normal' fifth harmonic (the interval between any two
notes of which frequency ratio is 5). This he divided into 25 equal
smaller intervals, so that each successive note on his scale is about
7% higher than its predecessor. In consequence, very few of the
intervals based on this scale correspond to anything in the musical
experience of the listener.
If music following this scale were played on a conventional instrument, almost any combination of notes attempting either harmony
or melody would be dissonant because of the harmonics present.
Instruments such as violins or trombones could in principle be
adapted to playing it; others, such as trumpets and woodwind instruments, would have to be specially made. But it still would not work,
because of the unwanted musical harmonics. The only feasible form
of construction for such a work was with completely synthesized
sound from electronic sources.
Stockhausen, having devised his novel scale, proceeded to put
together new timbres in which the partials would all fit in with each
other on the scale. He used just five basic sound groups, each of
five tones. The most compact of these contains five successive
frequencies (`notes') from the new scale. This consists of a series
of dissonant pairs, and its quality could be described as `astringent'.
The next group has members spaced two notes apart, and the internal
dissonance has now almost gone; in the other groups, with members
spaced at three-, four- and five-note intervals, it has gone completely, and the difference in character between them depends solely on
their width of spectrum. But if when groups are played together
there are strong adjacent pairs, dissonance reappears. This single
example is offered as a stimulus to the reader's imagination. Plainly,
the opportunities for shaping sound go far beyond what is conventionally achieved.
[17.9] Animated sound on an
optical film track
1±3, Cards from a set giving different
pitches. More complex patterns of
density give richer tone colours.
4±8, Masks (to be superimposed on
tone cards) giving dynamic envelopes
of a particular duration. Long
durations are obtained by adding
together cards that are open at
the ends.
9, A card giving a range of pitches of a
particular tone colour. A wide range
of similar techniques has been
employed, using variable density or
variable width to control volume.
Animated sound
Finally, in what may be a completely blind alley off this digression,
the following is an account of the (literal) use of shape to create new
sounds.
For almost as long as film has carried sound, optical tracks have
been used as a medium for `animated' sound. Drawings have been
photographed or marks inscribed directly on the soundtrack, either
frame by frame or continuously (printed by rotating wheels). Sometimes sound was constructed by filming geometric forms such as
280
Shaping sound
rectangles, triangles, ellipses, or lettering, fingerprints, even facial
profiles. Pitch was controlled coarsely by changing the size of the
shape, and finely by camera distance. Double exposure superimposed one sound on another. Light slits linked to pendulum systems
have also been used.
The Canadian animator Norman McLaren made no attempt to
imitate natural sounds. Instead, he showed that quite simple shapes
could be rich in harmonics, with some producing strident or harsh
effects. He sketched a range of `tone colours' for use with tone
contouring masks, superimposed cut-outs that had an attack, midnote dynamic contour and decay. He shot frame by frame, but
sometimes extended an image over several frames to produce a
continuous or mandolin quality. For a pause he used black card.
The resulting blips, clunks and splinges could sound as though they
were squirted in blobs or squeezed like toothpaste ± and the difficulty in describing them illustrates his success in getting away from
orthodox sound.
Chapter 18
Audio editing
At one time, audio editing almost always meant cutting and rejoining
a tape, and many of the terms that are still used refer directly to this
physical process. Tape, obviously, is a linear medium: to get to any
particular place on it, you must run forward or backward to that
point.
Now the same ends are generally achieved by reorganizing and
recombining data in computer memory. This begins with recording
and laying tracks on hard disk. Rough editing is assembling the main
body of a programme or sequence in the right order and taking out
the longer stretches of unwanted material. Fine editing is tightening
up this assembled material to provide for the best continuity of action
and argument, and also deleting hesitations, errors and irrelevancies
that may distract attention and reduce intelligibility. Finally, the
edited assembly will usually require mixing, in which several (or
many) tracks are combined in a final mix.
Hard disk editing is non-linear: Any point can be accessed on any
track in any order at any time. It is also, generally, non-destructive:
the original recordings are undisturbed. To edit in this way, you
make a series of decisions on what should be done, and the computer
demonstrates what would happen if these were carried out. It looks
(or sounds) just like the finished product, except that all you have at
this stage is a list of instructions for instant non-linear access. This is
called an edit decision list, or EDL. The EDL can be revised again
and again without affecting the original; neither is it immediately
destructive of earlier versions of the EDL. You can instantly `undo'
several (often many) stages of the edit. If you save an edit from time
to time, you can easily return to these even older versions. (And if
copies are saved safely somewhere else, you can protect yourself
from crashes.)
Tape editing is not the exact opposite of `non-destructive'. The
original would only be destroyed by wiping or recording over
282
Audio editing
deleted material, and although this linear editing process has been
used, that is not how it is usually done. Instead, the tape is cut, and
deleted material can in principle be joined up on to a second reel for
potential reinstatement or reuse. This, of course, is far more laborious than working on disk.
Sometimes a combination of the two processes is used. The main
edit is done on hard disk. The result is played out (allowing a final
check on the edit) and recorded on analogue tape as backup, prior to
broadcast.
For any simple last-minute changes, cutting the tape is fast and
convenient. And using tape rather than hard disk for transmission
protects the broadcast from crashes.
As computers become more reliable, or (a more likely future)
`broadcast-critical' access to independent backups becomes a safe
alternative, the use of tape will diminish further. Its own safety
features have helped it survive alongside its digital competitors,
but as broadcast play-out systems demonstrate that local computer
failures can be bypassed, the additional features that integrated
digital systems can offer ± including centralized control of every
aspect of content from commissioning and scheduling to a final
archive resting place, alongside a full account of copyright and
history ± will make them unbeatable.
Editing techniques will be described below in the order of their
historical development. But whatever technique is employed, editing
is generally undertaken for one or more of five reasons ± which
apply to television and film as well as audio for broadcast or
recordings:
.
.
.
.
.
Programme timing ± to adjust overall duration to fit some
scheduled space. In radio or television this often overrides all
others, whether or not the final stages of editing improves the
result ± although any original treatment should be written with
the intended length in mind.
Shaping the programme ± to give it a beginning, a middle and an
end, while ensuring that pace and tension are maintained. In
actuality speech recordings, any awkward parentheses, repetitions, redundant phrases or even whole paragraphs should go if
they obstruct flow.
Inserting retakes. This restores or improves a predetermined
structure, especially in music and drama.
Cutting fluffs, etc. Minor faults can lend character to unscripted
speech, but too many make it difficult to follow. Mistakes in
reading a script do not often sound like natural conversational
slips.
Convenience of assembly ± to combine location recordings with
studio links; to allow material of different types to be rehearsed
and recorded separately; or to pre-record awkward sections.
Radio and television techniques that aim to avoid editing presuppose
either that the broadcast has a well-rehearsed and perfectly timed
script or that those taking part are capable, in an unscripted
programme, of presenting their material logically, concisely and
Audio editing
283
coherently, or are under the guidance of a presenter who will ensure
that they do. Such self-control under the stress of a recording session
is rare except among experienced broadcasters. Even when a recording is made in familiar surroundings, the intrusion of the microphone
can produce stilted and uncharacteristic reactions. One purpose of
editing is to restore the illusion of reality, generally in a heightened
form.
Editorial selection from unscripted material often implies responsibility for avoiding misrepresentation of the character or intentions of
participants. On hearing an edited playback, any contributor expects
to recognize the essence of the original, within the limits of what
was needed for the programme, which should have been made clear
in advance. But editing is sometimes used to make content conform
more closely to the purposes or preconceptions of those in control of
the process. At best, this may be to mould it to some openly declared
format. At worst, it subverts the intentions of contributors. Plainly,
editing can cause controversy, and may lead to debate on ethics ±
which may overshadow the process, even as it is happening. But the
purpose here is to describe what can be done, and how. Fortunately,
in most cases, the purposes are both clear and desirable.
Retakes and wildtracks
Even as the original recordings are made, the first preparations for
editing begin. Nearly all recordings benefit from at least a little
tidying up, and this is easier if, right from the start, you constantly
check what might be needed. What to watch for depends on the type
of programme. For example, if a `fluff' occurs in speech, will it be
desirable or even possible to edit, or would a retake be advisable?
On film or during a video recording, the error may be at a point
when a picture cutaway is intended or can easily be arranged. In this
case a verbal repetition (for example) may be left for the editor to
cut, but if this is awkward a retake might be recorded in sound only
(wildtrack). If a retake in vision is necessary, the director has to
decide whether to go again from the start of the scene; whether
to change the framing of the picture (often, if possible, to a closer
shot or different angle); or whether the same framing can continue
after an earlier cutaway. The sound recordist should also (generally
without being asked) supply background atmosphere (`atmos' or
`buzz track') at the same level and in sufficient quantity to cover
any gaps that may be introduced by editing or where there are more
pictures than synchronous sound. This saves time and effort later, at
the re-recording (`dubbing') session.
In a radio discussion between several people, the director should
note whether the various voices have clearly identified themselves,
each other, and also what they are going to talk about. It might be
sensible to record a round of names and subject pointers in each
voice so these can be cut in if necessary. Watch for continuity of
284
Audio editing
mood: if the point a retake is likely to join up to ends with a laugh,
try to start the new take with something to laugh about (which can be
cut out). Discourage phrases like `as I've said', which all too often
refer back to something that will be cut, and retake if they, too, are
not clearly editable.
When editing radio programmes, `pauses' may be needed. So, here
again, record 15 or more seconds of atmosphere. If absolutely
necessary, atmosphere for pauses can often be picked up from
hesitations between words, but it is better to be safe. Record `atmos'
while the speakers are still there: after they have gone the sound may
be slightly different. If much editing is likely, be sparing in the
control of levels: over-control bumps the background atmosphere
up and down. If the background is heavy or varied, record extra or
longer tracks.
At a recording session for classical music it is again useful to record
a few seconds of atmosphere following each item or at the beginning
of the next in order to have something to join successive pieces.
Music retakes should be recorded as soon as possible after the
original, so that the performers can remember their intonations at
the link point, and also so that the physical conditions of the studio
are close to the original. An orchestra quite literally warms up; its
tone changes appreciably in the earlier part of each session in
a studio.
Tape editing
For many years the standard way to edit mono and then stereo sound
was to cut and splice analogue 6.35 mm (commonly known as
quarter-inch) tape. The tools required are the following:
.
.
.
[18.1] Equipment for cutting and
rejoining audiotape
1, Splicing block. 2, Blade.
3, Leader tape. 4, Wax pencil.
5, Jointing tape. 6, Spare spool.
A splicing block with a channel to hold the tape. The channel is
lapped slightly so that as tape is smoothed in it is gripped firmly.
The block has cutting grooves angled at 45 for mono, and 60
and 90 for stereo. It should be fixed firmly at the near edge of the
tape deck.
A fairly sharp stainless steel razor blade ± `stainless' is nonmagnetic. Use single-sided blades (and preferably not too sharp)
to avoid cutting fingertips. Store them safely in a holder.
A soft wax pencil (yellow or white) or felt-tipped pen, for marking
edit points on the tape backing.
Audio editing
.
.
285
A roll of splicing tape (non-oozing), a little narrower than recording tape to allow some inaccuracy in use without overlapping the
edges. (Pre-cut adhesive `tabs' in a dispenser are also available.)
Leader and spacer tapes ± white or green for the leader, yellow for
spacers and red (or twin coloured, red and white) to mark the tail.
A dusting powder such as French chalk may be used in hot climates,
where the adhesive may ooze a little. It is important that tape should
not be sticky, or layers will pull as they come off the spool, drag on
tape guides and clog the heads.
When preparing to make a cut, first play the tape to the point at
which the cut is to be made, and then switch off replay, so the tape
can be spooled by hand with the tape in contact with the head. (This
head-scrub effect is mimicked by most digital editing systems.)
Check the exact point by pulling the tape back and forth over the
replay head. Mark the tape on the backing, and always in the same
way, e.g. along the lower edge: this ensures that if a scrap of tape is
reinstated, it will go in the right way round. A simple edit is made
like this:
1. Locate and mark the joint.
2. Switch the recorder to a condition that allows the tape to be lifted
from the heads, and place it coated side down in the editing block.
3. Cut. The razor blade should be sharp enough not to drag when
drawn lightly along its guide channel. No pressure should be
necessary.
4. If a fairly long piece of tape is to be taken out, replace the take-up
spool by a `scrap' spool and wind off to the new `in' point. Mark
this, cut again, and swap back the take-up spools.
5. Place the two butt ends of the tape coated side down in the block,
a little to one side of the cutting groove, so they just touch
together.
6. Cut off a short length of splicing tape and lay it over the butt,
using one edge of the channel to steady and guide it into place.
7. Run a finger along the tape to press the splice firm. If it seems
sticky, dip a finger in French chalk and run it along the tape again.
[18.2] Audiotape editing
1, Move the tape backward and
forward over the heads, by hand, in
order to find the exact point. 2, Mark
the tape with a soft wax pencil. 3, Cut
the tape diagonally using the groove in
the block. 4, Butt the two wanted
sections together in the block and join
with special adhesive tape.
After a splice is made, place the tape back on the heads, and wind back
30 cm (1 ft) or so ± usually spooling by hand against the braking
effect of clutch plates ± then play through the splice, checking that:
.
.
.
.
.
.
.
.
it is properly made;
the edited tape makes sense ± you have edited at the point
intended;
the timing is right ± that the words after the joint do not follow
too quickly or too slowly on those before;
the speaker does not take two breaths;
neither the perspective nor volume of voice takes an unnatural
jump;
there is no impossible vocal contortion implied;
the tone of voice does not change abruptly, e.g. from audible
smile to earnest seriousness;
the background does not abruptly change in quality, nor is any
`effect' or murmured response sliced in two;
286
Audio editing
.
(for stereo edits) there is no `flicker' ± a momentary sideways
jump that can be introduced by an imperfect butt.
A virtue of this kind of editing is that it is easy to reinstate material if
the cut does not work, or to change it by cutting in a different place.
You may need to cut down a gap that is too long; insert a pause, a
breath or even an `um' or `er' to cover an abrupt change of subject or
mood; or you may have to restore part of what was cut.
[18.3] Angled butt joint in
audiotape
As the audiotape (held together by a
strip of jointing tape on the outer
surface) passes over a recording or
replay head, there should be no loss
of contact at any time, unless the butt
is imperfect. An angled joint
minimizes the effect of any slight gap
or overlap; it also creates a rapid mix
of outgoing and incoming sound. This
is suitable for monophonic
recordings, but not all stereo (the
position of sources in stereo can
momentarily `flicker' to one side).
To remake a splice, first mark any new point you propose to cut at, then
undo the old join. To do this turn it over and break it; grip one of the
points of the butt and pull it away from the splicing tape, then the
other. After remaking the cut, back it with a strip of new splicing tape.
It is recommended that monophonic tape (including mono contributions to a stereo mix) should be cut at an angle. If it is cut at 90 , any
sound on the tape, even atmosphere, starts `square' at the point of
cut, and any sound cutting in at full volume seems to start with a
click. But if a fade in or out exceeds 10 ms in duration there is no
click: in fact this is over 20 times more than is actually needed in
most circumstances. This duration corresponds to 2 mm of a 19 cm/s
(i.e. 0.075 in of 7 in/s) tape and works well for mono. For stereo an
angled cut (at 60 ) should be used only when editing in a pause or
between similar sounds, otherwise a momentary change of position
(flicker) will occur. If this seems possible, use a 90 cut (or a mix).
The techniques for rough and fine editing of content were developed
using analogue tape, and are equally valid for digital editing.
Rough editing
For this, the simplest and most flexible example is that in which only
sound is to be edited, typically for radio or records ± but much of what
follows applies equally to video editing for television, film or DVD.
When editing begins, it may be tempting to plunge straight in, to
work in detail on some accessible sequence ± but this might be a
mistake. A sequence that is cut so early often ends up with more
recuts than any other, as it is weighed and then reconsidered time
after time in the context of other sequences. It is generally wiser to
begin by putting together material that helps to outline the overall,
intended structure. In rough editing, the first task is simply to
assemble material in its intended order, at this stage including only
those retakes that contain essential highlights. The aim is to get an
early glimpse of the ultimate logical or artistic effect, without having
to spend much time or effort on the mechanics (or artistry) of
smoothly linking it all together.
Remember that that it is easy to transpose material. In radio, this can
be useful when unrelated subjects appear in an unscripted discussion. Or when editing a recorded `actuality' event, it may help to
explain what is going on if some of the material is taken out of
sequence. In any case, factors that govern the order of recording may
Audio editing
287
not hold for the finished programme. For example, it is often best to
start with a striking effect that engages the listener's attention and
involvement. Then the middle part of a work should, while progressing
logically, have light and shade: with variety of pace (in a discussion, of
speech lengths and rhythm) and of mood (or emotional development).
Ideally, the tension should build to an effective `curtain': in a discussion, something that contains in a few lines the essence of all that has
preceded it, or that will offset and illuminate the conclusion ± a `payoff' line.
As material is selected it is added to the master recording in
programme order, with transpositions and major retakes in their proper
places. At this stage use clean, simple cuts: in speech, cut mainly at
`paragraph pauses'. Anything that is not selected in this first pass is
retained (and kept readily available) until the work is complete.
Even at this early stage exercise care: try to avoid cutting sound in
the middle of a breath, or too close to the first word when there is a
heavy background. When joining different sections delivered by the
same speaker, check whether voice quality, mood and balance are
sufficiently similar: it is all too easy with radio for the listener to get
the impression that someone new has entered the conversation and to
wonder who that is, instead of listening to what is being said.
If subsequent mixing is envisaged, different components will eventually be laid on separate tracks. For a first, rough edit on quarterinch tape, sound from each source can (mostly) be laid sequentially
on a single track: if necessary, this can be split to two tracks later.
Fine editing
When the rough assembly has been made, give thought (again) to the
shape of the programme as a whole. It may now be clear that you
have to make severe internal cuts in some sections but less in others;
obvious cuts may include overlaps, repetitions or unnecessary
elaboration. For many types of programme, ranging from radio discussion to television documentary, you may also want to check the
intelligibility and conciseness of individual contributions, and
whether the character and personality of participants is adequately
represented. Try to keep the theme of what remains tight and unified.
Some other things that may need to be cut are the following (some of
these are easier for radio sound than where pictures complicate it):
.
.
Heavy coughs, etc. These hold up action and may cause the
audience to lose the thread. But if the speaker has a frog in the
throat (and this section cannot be deleted), it does no harm to
leave in the cough that clears it. The audience will feel more
comfortable too.
Excessive `ums' and `ers' (or any other repetitious mannerisms)
should be cut if that improves intelligibility, which it often does if
they awkwardly interrupt a sentence. But some actually improve
288
Audio editing
.
.
.
intelligibility by breaking the flow at a point where it needs to be
broken. So be careful: there are people who use these noises to
support or modify meaning, or to add emphasis or emotional
expression, and while the result may not read well, it sounds
right. Others seem to express character in these punctuating
sounds. In either of these cases, do not cut. An `er' moulded into
the preceding or following word often cannot be cut anyway.
Excessive pauses. In real life (or on television or film) we often
use pauses to examine and read a speaker's face. On radio, a
pause is just a gap, so is a strong candidate for trimming unless it
lends real dramatic emphasis to its context. But note that pauses
should not be cut to nothing if the effect is of impossible vocal
contortions. Try this: record the sentence `You (pause) are (pause)
a (pause) good (pause) editor' and then cut the pauses out completely. The result will probably sound like a case of hiccups. The
minimum gap depends on how much the mouth would have to
change in shape during the pause. This applies to `ers' too.
Superimpositions. Two people talking over each other can be
irritating. But edit with care: it may be necessary to leave some
overlap, or to take out part of the sentences before and after.
`Fluffs' (verbal errors), in cases where the speaker has gone back
over the words to correct them. Again, an edit generally improves
intelligibility. But take care here too: the first word of the repeat
is often overemphasized.
In either discussion or documentary, each different speaker presents
a new set of problems and decisions on whether to cut or not. For
example, an interview may begin hesitantly and gather speed and
interest as it progresses. This is wrong for the start of a new voice on
air, when the audience's attention and interest must be captured in
the first few words. If you cannot cut this `warm-up' stage completely, it may be necessary to tighten up the first sentence or two:
this can matter more than hesitations at a later stage, when voice and
personality have become established.
In certain cases, particularly where the recording is of importance as
a document, no editing should be done at all (with the possible
exception of cutting down really excessive pauses), and the greatest
care should be exercised if any condensing is to be attempted. If in
doubt, cut away to another, linking voice.
In any case, do not over-cut. It is useful to have good ideas about
where to cut, but it is just as important to know where not to cut.
Finding the exact point
When editing speech internally, there are good and bad places to
cut ± between or even right in the middle of words. Some suggestions
about how to make good edits better are given below.
In a `sentence' pause, there are two places where you can cut. One is
after the last word of the sentence (and any associated reverberation)
Audio editing
289
has finished but before the breath that follows; the other is after the
breath and before the next word. In most cases it is best, and safer, to
make the cut after the breath: you then retain the full natural pause
that the speaker allowed, and can cut as close as you like to the new
word, which helps to mask any slight change in atmosphere.
[18.4] Human speech
Analysed to show formant ranges.
1, Resonance bands. 2, Unvoiced
speech. 3, Vocal stop before hard `c'.
There would be a similar break before
a plosive `p'. 4, Voiced speech. These
formants are unrelated to the
fundamentals and harmonics, which
would be shown only by an analysis
into much finer frequency ranges
than have been used here. This
example is adapted from a voiceprint
of the author's voice. Each individual
has distinct characteristics.
Cutting between the words of a sentence is trickier. In addition, the
words themselves sound very different if they are slowed down, so
that a suitable point is more difficult to locate. But certain characteristics of speech can soon be recognized at slow speed, e.g. the `s' and
`f' sounds. The explosive consonants `p' and `b', as well as the
stopped `k', `t', `g' and `d', are easy to pick out (though not always
as easy to distinguish from each other) because of the slight break in
sound that precedes them. In a spoken sentence the words may be
run together to such an extent that the only break may be before such
a letter ± perhaps in the middle of a word.
Analysis of human speech showing how its frequency content
changes as a sentence is spoken makes these transitions very clear.
The `zoomed-in' screen display available on many digital editing
programs is not as clear as this, but with a little practice, sounds can
be matched to shapes, and tiny gaps can be spotted.
Do not assume that because a letter should be present in a spoken
word, it will be there. Whole vowel sounds may prove to be absent.
The personal pronoun `I' is typical of many such sounds. In the
sentence `Well I'm going now', it may be missing completely. If you
try to cut off the `well' you may even find the `m' was a vocal
illusion, no more than a slight distortion of the end of the `l' sound.
Similarly, in `it's a hot day', spoken quickly, there is probably no
complete word `hot'. Nor is it possible to isolate the word from `hot
tin roof', as the first `t' will be missing and the one that is there will
be joined to the following vowel sound. Before attempting fine
editing it is worth checking what the beginnings and ends of words
sound like at slow speed (and look like on screen), and where words
blend into each other and where they do not.
In cutting between words, cut as late in any pause as possible. When
searching for the right pause on a recording, do not be misled by the
tiny break in sound that often precedes the letter `t' and similar
consonants. You can often cut in to a continuous sound, for example
where several words have been run together, provided that it can be
`ramped' in over a tiny fraction of a second (the equivalent of what
when cutting audio tape would be an `angled' cut). Do not attempt to
cut out of a continuous sound, unless you join straight on to a sound
of equal value (one for which the mouth would have to be formed in
the same shape).
It is very difficult to insert missing `a's' and `the's'; these are nearly
always closely tied with the following word, so that unless the
speaker is enunciating each word clearly and separately, it is necessary to take this in combination with the sound that follows. The
letter `s', however, is somewhat easier to insert or (sometimes) remove.
These principles apply equally to digital and analogue editing. When
editing video (or film), the way in which cuts are made has to
290
Audio editing
be related to the frame interval, to ensure that words and picture
stay in sync.
Editing sound effects
Editing staccato effects and absolutely regular continuous effects
presents no problems. Effects that are continuous but progressively
changing in quality can be difficult to edit, as can speech accompanied by heavy, uneven effects or speech in heavily reverberant
acoustics.
Rhythmic effects on their own are fairly simple, as they can be
edited rather like music (see below). Speech in a reverberant acoustic
shares that quality with music, so it, too, can sometimes be edited in
the same way as music. With more uneven noises it may be difficult
to find two points that are sufficiently similar in quality for a simple
cut, so for these it is better to overlap and mix. For example, when
cutting a 10-second burst of applause to 5 seconds, a simple jump to
a later point would make a blip at the join, but a gentle overlap
sounds fine.
Sometimes a jump in the background behind a speech edit can be
covered by mixing in some similar effects afterwards. Even so, a
change of quality can often show through unless the added effects
are substantially heavier, or are peaked on the splice.
There are certain cases where the untidy effect of a succession of
sharp cuts in background noise can be tolerated. It may be accepted
as a convention that street interviews (vox pop) are cut together by
simple editing, with breaks in the background appearing as frank
evidence of the method of assembly.
Editing music
Reasons for editing music include: to cut out a repeat, or one verse
or chorus; to insert a retake; to delete everything up to a certain point
so that we hit a particular note `on the nose'; to condense or extend
mood music to match action, and so on. Most of the skill lies in
choosing and marking the right part of the right note, and this
requires practice.
The important thing in music editing is not that the note after each of
the two points marked should be the same, but that the previous one
should be ± that is, when cutting between two notes it is the quality
of the reverberation that must be preserved. Any retake should start
before the point at which the cut is to be made: there is nothing so
painfully obvious as a music edit where a `cold' retake has been
edited in, resulting in a clip on the reverberation to the previous note.
It might be supposed that if the new sound is loud enough the loss
Audio editing
291
may be disguised, but this is not so unless it is cut so tightly as to
trim away part of the start of the note. This attack transient is
particularly important to the character of a sound. (AR could probably be used to disguise a bad cut, but this is a messy way of
working.)
In general, unless the cut is made at the end of a pause that is longer
than the reverberation time, the first requirement for a successful
music edit is that the volume and timbre preceding both the `out' and
`in' points on the original tape should be identical.
It may be possible to make an exception to this rule when cutting
mood music to fit behind speech and effects, but only if the level of
the music can drop low enough to `lose' the transition. For preference, choose a rest to cut out on and something rather vague to come
back to, and if necessary dip the music level at the edit point.
However, no such liberties may be taken with foreground music.
For example, when linking music is being trimmed to length, any
internal cuts must be made with the same care as would be applied to
featured music. Speech that has been recorded with music in the
background can be almost impossible to edit neatly, perhaps leaving
an untidy join in the music to coincide with a slight gap in the
speech.
When everything up to a certain point in a piece of music is cut in
order to create a new starting-point ± e.g. to come in on a theme ± the
effect of any tail of reverberation under the new first note is often
small. Except when this tail is loud, and contrasts strongly in quality
with the first wanted sound, it will be accepted by the ear as a part of
its timbre. Even then, a quick fade-in may sound better than a cut. To
get out of a piece of music at a point before the end where there is no
pause, there is no alternative to a fade. When editing on hard disk (or
any system where events are controlled by time code), the manual
fade that might be used on a live show can be replaced by a
programmed fade (a ramp).
When the cut is to be made on tape at a place that is not clearly
defined by the music itself, it is often possible to identify and match
the cutting points by either timing the distance on the track from a
more easily located point, or marking out the rhythmic structure of
the music. Of the two methods, the second is the better guide, as the
easily located note may not have been played exactly on the beat. On
audiotape the lengths are marked out on the back of the tape with a
wax pencil, so that the edit point can be calculated and marked for a
physical cut; in digital editing it is calculated by time code.
[18.5] Early hard disk editing
equipment
AudioFile system layout: 1, Controls
(`control surface') and VDU.
2, Equipment rack. 3, Audio inputs.
4, Eight independent audio outputs
(allowing the system to operate like a
cartridge machine). 5, DAT recorder.
Hard disk editing options
The time code generally used to locate and control audio material
was originally designed by the SPMTE (Society of Motion Picture
and Television Engineers) to work with picture and sound together.
It is usually displayed in an eight-figure format: hours, minutes,
292
Audio editing
[18.6] AudioFile editing system:
control surface
1, `Soft key' function switches.
2, Soft key status displays on VDU.
3, Selected page of information: for
example, see below. 4, QWERTY
keyboard. 5, Numeric keyboard.
6, Trigger keys. 7, Floppy disk drive.
8, Transport, nudge keys, `soft
wheels' and LED display for locating
and shifting material.
seconds and frames (for American video that is 30 frames per
second, but in Europe it is 25). Two more subframe figures may
be added, based on one-eightieth of a frame (or sometimes onehundredth). This allows digital audio editing to be flexible to 0.0005
of a second. Time code is described further, in connection with video
and film editing, in Chapter 19.
Today, the most common systems for editing on hard disk are those
that use the processing power of a host computer. But this is just one
of several ways in which it can be organized. Among the earliest
were dedicated systems in which the hardware (including all the
controls and displays) was built around the software.
[18.7] AudioFile control screen
`Audio record': a typical page shown
on VDU. Many have the eight tracks
of sound, which `play' across the
screen from right to left, past the
vertical line that represents `now'.
There may be many more tracks that
are not shown. Positions are based
on time codes, but can be measured
to greater precision. `Audio record' is
the only page that has a meter
display.
In one, the control surface has a keyboard, function keys, two scrub
wheels and a floppy disk drive, alongside and below a display screen
set in a small panel with more keys and buttons around it. The
display has switchable `pages' that allow the editor to operate in a
range of different modes. The sound must first be loaded (digitized).
If the sound comes from a source that has no original time code, this
must be added, in order to locate and control it. In any such system
the precision and repeatability of timings, programmed levels (to a
quarter of a decibel), cuts and mixes are far greater than can be
expected from manual control.
In this system, the screen shows 16 of the tracks (of up to 24 in
memory) in a display that looks like a multitrack tape marching from
left to right across screen, past replay heads in a vertical line down
the middle. Each track is identified by number, and a suffix, 1 or 2,
indicates the left or right track of stereo. A cut in or out is shown by a
vertical bar across the track, which travels with it; a fade (here also
called a ramp) by a diagonal. Sound can be bumped from one track
to another instantaneously. Just as easy, but less obvious once it has
been carried out, is a shift or slip to the side, moving a track (or part
of it) backward or forward in time. With minor variations, something
similar to this arrangement appears on all multitrack editing displays.
In another, more modestly scaled desktop editor that works with a
16-bit linear (i.e. uncompressed) signal, the hardware looks more
like a standalone word processor, but with dedicated control keys
and a scrub wheel alongside the keyboard. Its small window displays
several tracks or can zoom in to the fine detail of one of a single
waveform.
Audio editing
293
In hardware of this kind, there are transport keys (similar to those on
a tape machine) to record, play, pause, stop, fast-forward and
rewind. There are editing keys to cut, copy, insert, erase, to locate
and mark editing points or to repeatedly play a segment as a loop, all
of which are non-destructive, i.e. the original recording is preserved.
There are also keys to save work, zoom in on the display, or undo
previous edits.
[18.8] Short/cut desktop hard disk
recorder and editor
Intended as a successor to earlier
generations of analogue and digital
tape machines, this 44.1 or 48 kHz
16-bit linear (i.e. uncompressed)
recorder is designed for rapid editing
and replay of near-live news stories
and phone calls. The scrub wheel
simulates the movement of tape over
a replay head. Editing (including
volume control, fades and crossfades) is rapid and precise. It holds 12
hours of stereo, with 10 clips ready
for instant playback: an external Zip
(or Jaz) drive can be used for archive.
[18.9] Tascam hard disk 24-track
recorder and editor
Designed to replace both analogue
and digital multitrack tape recorders
for music recording (24-bit and up to
96 kHz) but with similar transport
controls, including a scrub wheel,
plus time code, MIDI machine control
tools, DVD mastering, and a variety of
analogue and digital input and output
options. A wide range of music
editing operations, including real-time
cross-fades, are also controlled from
the front panel.
This desktop editor is designed as a simple (and speedy) advance on
tape editing, but with added features such as cross-fade, fade in or
out, gradual rise or fall in level, etc. The internal hard disk holds
over 11 hours of stereo. For storage of the edited material, the
makers recommend an external Zip disk, or for longer recordings a
Jaz disk.
Another hardware-based system is built around higher quality 24-bit,
24-track hard disk recorders, with a separate control surface to
display and edit the tracks, which may come from several recorders,
stacked to work together.
Several editing applications are designed to operate on a standard
office or home computer, controlled either from its mouse and keyboard
or more conveniently from a separate control surface that has all the
usual dedicated keys, knobs, faders and scrub wheel ± although the
keyboard is still the fastest and easiest way of typing in titles and
other notes. Without the hardware, the keyboard's function and other
keys are assigned as `hot key' transport and edit controls (for which
it is convenient to have a mask that fits over the keyboard, so that the
function of each key can be marked). But operating faders and rotary
knobs by mouse from the screen is inevitably fussier.
It will also often be necessary to have a convenient way of moving a
project from one machine to another, from studio to office or home
and back, possibly to another studio. One device used for this is a
removable hard drive that can be connected by SCSI (from which
the `scuzzy' box takes its name). The hard drive ± somewhat
exposed when not plugged in ± is carried around in its own miniature
protective case. Another essential is a means of backing up the
original recording and other material in the studio: a DAT, multitrack or some other recorder where it can be kept safely while work
on it continues elsewhere.
Digital projects and editing
Digital editing performs many of the same functions as those originally developed for audiotape or separate magnetic sound for film, so
the same general principles carry through, with exactly the same
stages of rough and fine editing, and the same need to locate and
mark editing points in speech and music. For hard disk digital
editing, the display on the screen mimics tapes passing replay heads,
with the edits clearly visible ± but with the added advantages that
other information can be superimposed, that a timeline and other
294
Audio editing
timings are displayed, that access is far more rapid, and that most of
the operations are (usually) non-destructive, instantly playable and
can be readily reversed. Most of the controls and procedures are
directed towards presenting and working on this pictorial representation of the audio plus any MIDI tracks (slightly different in layout)
that may be lined up with them.
[18.10] Digital audio edit station
1, Screen and loudspeakers.
2, Headphones as an alternative to
loudspeakers in an active studio.
3, Hardware controllers (optional) add
to the cost but are faster, easier and
more intuitive to use than a keyboard.
4, Keyboard: more convenient than
the hardware for naming. 5, Mouse:
used for drag and drop operations. If
there is no hardware controller, the
keyboard is set up with hot keys.
Typically, F1±F12 are used for editing,
and the numerals 1±10 in the next
row for mixer automation. The arrow
keys are used to zoom in or out and
move clips up and down, while the
keys above the arrows open the
clipstore or copy material to it, move
the playhead to the start or end of the
EDL, or save the whole project. The
descriptions here are for SADiE
editing.
The raw materials for editing on hard disk exist within the computer
as components of a substantial file that for each separate work may
be described collectively as the project. The project might be a radio
programme or series, a record or single element or package, and
comprises all the edit decision lists, clipstores and mixes that are
created alongside some original audio. It may also include `metadata', consisting of history, copyrights or even scripts. All this exists as
a single named entity within the computer, and can be copied,
perhaps on one `project backup' keystroke, to an external store ±
still with the original audio. As ever, vital questions are `How stable
is the edit program?' and (when it does crash) `How easy is it to
recover?' and `Will it let you carry on from where you left off?'
The following describes a typical, well-organized operational procedure: programs differ in detail, but the objectives and order they
are tackled in will mostly be similar, and there will usually be a
series of things to organize before the interesting part ± the actual
editing ± can begin.
As with any other computer application, a digital editor is started by
opening the program, often from an icon that it is convenient to keep
on the desktop. Completing any required log-in primes a start-up
procedure to open either the current or some other existing project,
retrieve a project from backup, or create (and name) a new project.
On creating a new project (which will open up a window for it) the
next stages are to record into it and set up an EDL (edit decision list).
The main features in the window (or that can be reached from it) are:
AES31-3 is a public-domain
audiofile interchange standard
that specifies how random
access audio can be described
and organized in order to
transfer simple audio projects
(with an unlimited number of
channels) accurately between
editing systems. It instantly
autoconforms precise audio in
and out times, together with
linear or shaped fades, plus
sources, clip names and
remarks. It is also designed to
include dynamic level (gain)
automation, but not EQ, filter,
limiter or compression data, as
these can sound different on
different platforms. It is used
on many (but not all) proprietary
systems. A more complex
alternative is OMF.
.
.
.
.
.
A number of boxes, tracks or `streams' across the screen (rather
like strips of audio tape) that will contain the audio, with (above
these) a timeline, and from top to bottom a line (equivalent to a
playhead across all the tracks) that represents `now'.
In boxes to the left of the streams, a `playlist' in which they are
listed by number (clicking here selects a stream or can be used to
combine two adjacent tracks as a stereo pair); there is also a set of
small icons that operate on each individual stream, and one of
those is a button that will prime it as a target for recording to.
A large number of icons that change the screen in various ways,
edit the playlist, operate audio editing tools, save the work, and
so on.
A mixer window that can be extended or moved around the
screen; faders, side by side, are numbered to match the streams.
A transport control window, with buttons like the controls of a
tape machine, plus time code to show where it is up to, and
naming tools.
To prepare to record, press a tab marked `record' near the top of the
transport window, then fill in any details required below: one will be
Audio editing
295
a name for the recording. It may also be necessary to bring up a setup window and choose the type of source: what kind of digital data it
is (e.g. AES3) or if it is analogue to be digitized (at some specified
sample rate, e.g. 44.1 or 48 kHz). To check that all is ready, try
pressing the `pause' button: a red `record' button may flash. Then
play in some material (or feed sound from the studio) and check that
the meter alongside or above the fader shows a signal (which might
just reach to the red). If there is a problem at this stage, recheck the
routing. If all is well, press the red `record' button to begin, and
when finished the `stop'. Finally, release the primer button on the
playlist.
To begin editing, press the `edit' tab near the top of the transport
control panel. Some systems allow editing to begin even while
recording (in `background') is in progress ± also useful for `bleepediting' offensive material out of `near-live' telephoned contributions.
[18.11] SADiE transport controls
1, By clicking on `Record', the lower
part of the window is configured for
information about the track. Release
the key after recording is complete.
2, Clicking on `Edit' configures it to
move selected material ± e.g. with
the scrub wheel (to hear this, click
a button near it). To activate the
mouse for drag and drop, click on
the mouse key in the corner.
The most useful display for the audio is as a waveform (there may be
a waveform icon: click on this first). Clicking on a clip should select
it for editing: it should then be possible to move the in or out points
while keeping the audio stationary, or move audio relative to a fixed
in or out point, or again, move the whole clip over the `playhead'
line. Clicking on a razor blade icon will cut a track at a point marked
by the cursor. All edits are provisional: it is easy to recall the original
material and use a scrub wheel to perfect the position, or to add back
a length that had been taken out. It is usually possible to mark a
section, then clip and paste it somewhere else, perhaps in another
track, or simply drag and drop it (just as in word processing). It
should also be possible to mark and move regions containing several
tracks together.
Colours may be used to show the status of each audio clip. Typically, from an original pale blue for mono or green for stereo, when a
section between two markers is selected prior to editing it becomes
dark red. Bright red shows `edit in progress', in which state it can
be extended forward or back to add more of the original. Dark blue
marks a clip that will `slip' to fill up (or open out) a gap.
In addition to its waveform, each segment between two marks may
show the name originally given to it in the top left corner of the clip.
In a short segment only a little of this name may be visible, but more
will be revealed if the waveform is expanded by `zooming in' on it.
If it helps, the name may be changed to give more information about
its source as speech, music or effects. Zooming in vertically at the
same time deepens the waveform, making it easier to locate a precise
edit point. Zoom in far enough and on a linear (uncompressed)
recording the waveform may be seen as a continuous line. (A compressed `waveform' may appear as a blocky histogram showing the
heights of successive samples.) Zooming out may eventually reveal
the coarse structure of an entire recording or edit, with quiet and loud
passages showing up.
The settings for volume control may also be shown superimposed as
a line over the waveform, dropping below it as the sound is faded
out. If the start and end of each fade can be separately marked, the
296
Audio editing
fades will `ramp' smoothly. To change the contour of a fade marked
like this, it is necessary to change volume over several marks. Cuts
are often made a little `soft', to avoid a click. Other operations on the
streams (or channels) are carried out much as they would be on a
console, except that controls (e.g. for EQ or AR) will appear in a
window on the screen, as do other DSP plug-ins that may be available. There will also be buttons to send the output from each channel
to a choice of buses. Many of these operations feel more familiar if
an (optional) external control surface is used.
[18.12] SADiE edit screen
This shows audio streams set up for
editing. Below left is a mixer: each
fader (with input and output buttons
above them) operates on the track
with the same number. Below centre
is the transport control window.
Below right is the project playlist
window. Other windows (such as
DSPs) can be opened over this
screen.
One disadvantage of some digital editing systems is that from
analogue sources they must be loaded (i.e. material is recorded in) or
played out as analogue in real time. It is therefore necessary to plan
for this to be done, if possible, at a time when it will not hold up
operations. One use for this time is to check the input or to review an
edit. In contrast, a CD-R can be loaded at perhaps 20 times normal speed.
Play-out systems
As an extension of the editing process, some systems can operate as
a play-out controller for items either in an active studio or in a
digitally automated broadcast system. Edited items are assigned to
a playlist, which works rather like a cart stack, except that it is no
longer necessary to carry around and load a pile of tapes, cassettes
or cartridges. It is also more flexible, in that it is easier to build,
edit components, and sign them off as ready for transmission, but
still make late changes or switch the order (either in an office
environment or at the back of the studio, or at the console). A
playlist compiled at the back of an active studio may be `mirrored'
(i.e. shown in another place) at the console, and used to select or
play inserts out in order.
Play-out systems work from playlists that automatically shift to the
next item as each is completed. It may be arranged that successive
items go to alternate channels (e.g. `A' and `B') so that an overlap
or cross-fade can link them. (If deletion of an unplayed item automatically shifts the channels of later items it is safer to keep both the
A and B faders open and control them together as a group. A
separate `C' list of frequently used shorter items might also be
played in from a source such as a Zip or MiniDisc cart.
[18.13] Play-out lists
For broadcast studio use, this version
has three columns. A and B, Alternate
ready-to-air segments can be
overlapped or cross-faded. C, Station
ID, signature music, trailers and other
frequently-used items that may be
interpolated into the schedule.
The playlist may be set up as a hierarchical `tree', with at one level a
whole day listed as hourly segments with station identification, etc.,
at the junctions. The segments may then be opened to reveal their
detailed structure with individual and progressive timings, and notes
on the content. The items on this list may be audio (such as records,
recorded links or completed pre-recorded items), placemarkers for
material yet to be selected or completed (with estimated timings), or
cues for live links, time-checks, or other notes. An item towards the
conclusion of a segment may be of flexible duration, perhaps prefaded, i.e. set to end, rather than to start, at a given time.
Audio editing
[18.14] Cart wall
A touch-sensitive (or mouse-keyed)
screen with 49 boxes each holding an
item that is ready for instant replay: a
two-layer hierarchy of such screens
allows rapid access to 2401 (49 49)
boxes.
297
Broadcast automation systems have been around a long time,
originally using carts of cartridges or cassettes, and then MiniDisc
or digital files on, for example, Zip disks. The cart may now be
replaced by a cart wall, a touch-sensitive screen with a 7 7 array of
boxes, to which items are assigned from a database, then replayed
instantly as each box in turn is touched. This could be used directly
to air, or to assemble playlists ready for broadcast. A valuable
adjunct to this technology is that each item can, in principle, be
accompanied by metadata that includes all its copyright data and
history, which can be continuously updated, then used to `report' the
copyright details and other content automatically after transmission
± so that fees due are also calculated (and payment can be set up)
without paper. All this assumes that all the different connected
devices can be linked by a standard metadata exchange format ± or
(to add yet another acronym) `SMEF'.
Broadcast automation in this form is, in principle, capable of managing
far wider swathes of broadcasting than just play-out. The seemingly
successive processes of commissioning, making, scheduling and
transmitting material, then `reporting' its use and archiving it, plus
multimedia offshoots of teletext and pictures and web data, all become
elements with a single interlinked process of data management ± a
big single `pot' of data with a set of tools attached. There is, of
course, one other ingredient that must be added to this pot: far higher
`broadcast-critical' reliability than individual computers can offer.
If even just a playlist is set up as a broadcast schedule, this should be
isolated from other uses that could accidentally disturb the order,
timing or content of transmissions ± `You wouldn't want your audio
out there fighting someone's Friday night Excel spreadsheet'. The
system must have proven reliability and redundancy with immediate
switchover in the event of failure. It should have procedures in place
to protect it from viruses, hacking, deliberate misuse and human
error or other ills that may afflict computers. Of several possible
arrangements, one has a central server, with a second comparable
backup server in another room, but ready to switch in instantly.
Another uses multiple similarly paired smaller servers, linked
to pairs in other locations, so backup is always available from
unaffected parts of the network, and any individual server failure
would be on a smaller scale.
Chapter 19
Film and video sound
This chapter offers a compact account of the principles involved in
video sound recording and editing, relating this to current television
and film industry practice. Note here that as a medium, `film' now
refers more to the distribution system that still uses it, rather than
production, in which the use of film is increasingly replaced by
video cameras and digital editing. Sound, in either case, is generally
recorded and edited digitally. This chapter deals mainly with picture
and sound that originates on location (including what is still called a
film set); Chapter 20 deals more with sound in the television studio.
Video cameras that are equipped to record picture can always record
audio alongside it. For sound this is plainly an economical approach,
with the added advantage that there are fewer synchronization problems ± at least, not at this stage. It is therefore suitable for television
applications such as newsgathering, for which speed, cost and technical simplicity are important. But just how satisfactory that `simplicity' really is depends on the position of the microphone. If the
microphone is on the camera, even though its output can easily be
fed directly to a recorder on the camera, this is likely to be unsatisfactory for two reasons: one is that the camera is very rarely the
best place for a microphone; the other is that it treats care and
attention to the quality of sound recording as an option, not as an
essential.
[19.1] Four-channel digital tape
recorder
A helical scan recorder, like DAT but
with a 6.3 mm (1/4 in) digital tape.
Records on four channels for 1 or 2
hours, depending on size of spools.
A better alternative, still with the recorder on the camera, is to use a
separate microphone, under the control of a recordist who feeds its
output by line or radio link to the camera. Again, synchronization is
automatic ± unless the recordist makes a backup, in which case it
should have the same time code on it as that on the camera recording.
For more complex operations, sound recordists prefer to use separate
recorders of established quality under their direct control, using the
link with the camera primarily for synchronization. Digital audio
tape (DAT) has been widely used, especially for television, but in
Film and video sound
299
the film industry, where high definition (HD) video is taking over
from traditional film cameras, digital media with four or sometimes
more tracks are increasingly employed (exceptionally, 24 tracks),
and in these cases any sound fed back to the camera recorder is for
guide and viewing purposes only. Other options include two channels on a removable 2.2 GB hard disk cartridge. And an interesting
extra facility with one DVD recorder is hard disk backup with up to
10 seconds of sound recorded from before the `start' button is
pressed.
Time code
[19.2] Removable hard disk
cartridge
A 2.2 GB hard disk holds over 2 hours
of high quality (24-bit) stereo.
A time code, as applied to video, is an eight- or ten-character number
indicating hours, minutes, seconds, frames and subframes. It can be
linked to a 24-hour clock, and so reset automatically to zero at
midnight. Alternatively it can be started from an arbitrary positive
number (such as 01 00 00 00, which means `one hour exactly'), after
which the numbers will continue to rise in real time as the machine
runs but will hold while it is switched off, then continue in sequence
when it is run again: this therefore measures total recording (and so
replay) time. If the tape starts at a number sufficiently far above
zero, it can park at a point before the nominal start and be ready to
run up without ever encountering negative numbers.
As an alternative to elapsed time, a recording can be marked with
`time of day', which is safest and clearest when several recordings
are made, and it also means that sequences separately recorded
over a long period can be identified more easily when consulting
a log.
[19.3] BITC
Burned-in time code offers eight
numbers for time and eight as `user
bits' (sometimes available for a
second time display). The position on
screen is arbitrary. On professional
equipment the data is kept separate
and combined, on demand, only on
the screen display. The combined
signal can be fed to a domesticquality recorder, where it will then be
permanently superimposed, and can
then be viewed away from studio or
cutting room.
The first form of time code was linear or longitudinal (LTC). This is
recorded continuously (originally on an audio cue track, where it
made a high-pitched buzz if replayed as audio). It can be `read'
when a tape is played or spooled, but this breaks down at low speeds.
On videotape, time code provides a method of individually marking
every frame and for this LTC is generally supplemented by vertical
interval time code (VITC), which is recorded intermittently in the
video signal just above the top of the picture; VITC still replays at
low speeds or even when the tape is on `pause'. Either can be used to
generate characters that feed number displays on the equipment
itself or superimposed on the picture as burned-in time code (BITC).
A video system with time code has separate feeds, with and without
BITC. If BITC is re-recorded on a domestic VCR, it permanently
overlays part of the picture.
Different time codes are used for film (24 fps ± frames per second),
European television (25 fps) and American television (30 fps or an
offset `drop frame' version of it that runs at 29.976 fps ± often
written as `29.97' or `29.98').
The digital numbers that contain the time code have some spare
capacity, called user bits, that can be used to assign tape numbers.
300
Film and video sound
Time code for audio clips
Because audio clips can be cut
anywhere, their time codes
must contain more information
than those for video or film.
AES31-3 uses the sampleaccurate Timecode Character
Format (TCF), which may
appear as HH:MM:SS:FF/1234.
Over a 24-hour clock, this
shows
hours,
minutes,
seconds and frames. It allows
for the various video or film
formats, including 'dropframe',
in which a second is divided
into 1001 parts. It is also used,
typically with 25 notional
'frames' per second, for soundonly editing. The 'slash' divider
signifies that in this example a
48k sample rate is being used
and the last four figures count
all the remaining samples
beyond the last full frame, in
order to define a cutting point
absolutely. Audio Decision Lists
(ADLs) using this format are
used to transfer data rapidly
and accurately between a wide
variety of proprietary editing
systems.
It may be arranged that at the press of a button the real-time flow of
numbers pauses, so that the time of a particular event can be logged.
When the flow resumes it jumps on, to the continuing real-time
display. In this way logged material can be readily relocated. More
importantly, however, it also means that computer-controlled operations can be initiated and related to any given frame on a tape. For
tape replay any required time code can be keyed in and the control
system set to search for that address. Once found, the tape automatically sets itself back and parks, ready to run up and display the
required material.
Digital video editing systems again lead to the generation of timecoded edit decision lists (EDLs), which are used to control sound
and picture recordings both for the mix and in a final conforming
process.
Time code can also be used for time-cued play-out in automated
broadcasting systems. It also synchronizes video with multitrack
tape or digitally recorded sound files in order to locate separate
soundtracks (e.g. the commentaries made for different users of a
common picture of some event or sporting fixture), and television
with stereo sound that is broadcast simultaneously by radio. In
addition, it provides an accurate measure of duration.
Picture and sound standards
The sound recordist may be faced with any of a wide range of
cameras and frame rates. Originally, film was used by television
(mainly 16 mm) as well as by the film industry (with 35 mm as
standard). In television, film is now rarely used (except where its
`special' picture quality is sought): it was replaced first by analogue
videotape, and this has given way to digital formats. The principal
format is now what is called standard definition (SD) digital, and
this is occasionally backed up by a more compact, lightweight DV
CAM format. For film industry purposes, 35 mm cameras are no
longer manufactured (but many directors still want to use them,
despite the high cost of film stock); increasingly the high definition
(HD) video format is replacing it. So far, this leads to three main
picture definition standards (there are others, which are mostly
variations on them and which will be disregarded here ± as will
the heated debates on the virtues of film compared to video).
[19.4] Digital video picture areas
compared
A, Standard video: 720 pixels wide,
678 lines deep (covers PAL and
NTSC formats).
B, DTV format: 1280 720.
C, HDTV common image format
(CIF): 1920 1080.
Unfortunately, other developments have led to a proliferation of
standards even for the camera types already in use. An important
question, which the sound department is unlikely to have much input
on, is choice of frame rate. The simplest case is for television in
countries where the electricity supply frequency is 50 Hz and 25 fps
has been adopted as the standard for television. Each frame is
scanned twice, once for the odd-numbered lines, then for the
even numbers and the two `interlaced' fields make up what the eye
sees as a single complete frame. This is now called `50i' (`i' for
Film and video sound
[19.5] Frame rates
Interlaced (television format):
50i
50 fields/s (PAL)
60i
60 fields/s
59.94i
NTSC offset
Progressive scan (like film):
24P
24 fps standard
23.98P
24 fps offset
25P
Euro TV standard
30P
American, 60 Hz
29.97P
Offset for NTSC
Choices available on a high definition
camera.
301
interlaced) ± and the leading picture and sound format using it is PAL.
Low-budget features may also be shot at this frame speed, so their
transfer to television is easy (although adjustments must be made if
they achieve cinema distribution). As it happens, DAT recorders were
also originally designed to operate at the European 25 fps.
In America and other countries that have the 60 Hz mains frequency
television, this becomes 60 interlaced (`60i') fields per second,
which makes 30 fps. The NTSC `off-set' variation on this actually
runs at 29.97 fps and the interlace is `59.95i'.
The film industry had already fixed on 24 frames per second, and for
the HD video version of this now prefers a picture that is scanned
just once, from top to bottom: this is 24P (the `P' is for `progressive
scan'). HD cameras are now available to operate at a wide range of
scan and frame-rate combinations.
There is an established process for converting between the 24 and
30 fps formats, and a slight difference in playing speed converts
30 to 29.7: the sound drops in pitch by a virtually insignificant
0.1%. Running 24P at 25 fps or vice versa changes both action
and the rate of delivery of dialogue by 4.1%. Television audiences
experience this regularly but audiences do not seem to mind, even
though a rise in pitch by about two-thirds of a semitone shifts
male voice formants a little closer to the female range, and music
is lifted almost into a different key. This effect can be corrected
by a digital signal processor set to control pitch and duration
independently.
There have been several, successive analogue videotape recording
standards. At one time television pictures and sound were recorded
on bulky and very noisy 2-inch tape machines (the Imperial measure
of width has always been used to describe this format). Special
transfer machines are now required for replay of any material that
has to be retrieved from 2-inch tape. After this came 1-inch, also
retained now only for archive material that has not yet been
re-recorded, and 3/4-inch, now used mainly for domestic or viewing
purposes.
For professional digital recordings 1/2-inch (and, on some cameras,
narrower) metal tape is used. Because most errors on digital recordings
are automatically identified and corrected, dropout has less effect on
them, and uncompressed data suffers no significant loss of quality
over several generations of copying. The number of channels of
sound has increased from one, then two, to four (or even eight), to
include a main stereo pair ± most users preferring M and S ± in
addition to any tracks for a microphone on the camera (or elsewhere), as well as the essential time code.
The analogue Betacam SP camera, once widely used for documentary and news, has four audio channels, but only two are linear (i.e.
along the length of the track and independent of picture), while two
are coded into, and so combined with, the picture recordings. Stereo
recordings are generally taken on linear tracks (1 and 2) and copied
to the coded tracks 3 and 4, but other formats are possible and should
be indicated in the documentation. Later systems are more versatile
302
Film and video sound
if all four (or more) audio tracks are independent of picture: there are
fewer problems when the same tapes are used for editing. All
professional video recording systems also have recorded time code.
More compact formats include plug-in hard disk cartridge.
Sound is synchronized with picture either by a physical link (a synclead) or by radio link. Another system that has been used for film is a
pair of crystals, matched to vibrate at the same frequency, that run
recorder and camera at the same speed.
In the rare cases where television material is still shot on film (for its
`special' picture quality), it is almost always immediately transferred
to video for digital post-production, after which it stays in the
television video format, ready for transmission. Many of the basic
principles for editing picture and sound together were originally
developed for use with film, but have now been adapted to the video
editing techniques that are used even when film is the original
source.
Digital editing systems are almost universal, with videotape (copy)
editing still being used only in the simplest cases. At one time, a
professional standard of digital editing could be achieved only on
specialized equipment, but for all but the most complex work this is
now challenged by software that can be run on a domestic computer,
although the more memory the computer has the better.
This chapter ends with a brief account of flatbed film editing, which
is still occasionally used. The techniques described apply equally to
16 and 35 mm, but where slight differences exist the method for
16 mm is described. Film (as well as film sound) is most conveniently edited using guillotine tape joiners and viewed on equipment
that allows continuous movement of picture, stabilizing the image
by the use of rotating prisms. This allows good-quality sound to be
played and heard in sync without the loud machine noise of old
intermittent viewers.
But this account must begin at the place where the combination of
picture and sound first appears. This is often far from a studio (in
which sound could, in principle, be controlled with relative ease), at
some location where the first requirement all too often has to be for
good pictures, while the achievement of even acceptable sound may
have to begin with an exercise in damage control.
Choice of location
When a location is chosen for video or film shooting, it is essential
to check its sound qualities and potential for intrusive noise. At
worst, bad location sound can make it necessary to replace dialogue
and effects completely: one branch of the film industry specializes in
this. But plainly it is simpler to record and use the original location
sound if this is possible. For documentaries this is almost always
preferred, not least for reasons of cost.
Film and video sound
303
Location surveys should establish that aircraft are not likely to
interrupt sound recordings. Direct observation is a good start, but
in addition a glance at an air traffic control map will show what type
of disturbance there might be ± passenger aircraft along main airways and in terminal control zones, and military or light aircraft
elsewhere. The survey should also ensure that road traffic noise is
not obtrusive (or, if traffic is accepted as a visual element, that it is
not excessive), and that machinery noise from factories, building
sites or road works will not create difficulties, and so on. If the
investment in good sound is high, a visit to the site at the same time
of day in the appropriate part of the week is a useful precaution.
In many cases where people unconnected with the production are
(in the course of their own normal activities) making noises that mar
a performance, a courteous request to stop often produces results. If
so, courtesy also demands that the noisemaker should be told when
shooting is complete or when there are long gaps. It is, of course,
unreasonable to expect people to stop earning their living in order
that you may earn yours, so in such cases (or where it is the only way
of gaining silence) payment may have to be offered.
For interiors, acoustics as well as potential noise should be surveyed.
On this will depend the choice of microphones, type of balance, and
remedial measures against deficiencies in the acoustic environment,
followed by debate (and often compromise) with people responsible
for camera, lights and design and, where necessary, recommendations to the director on how sound coverage could be improved.
Film and video sound recording
On location a video or film unit includes a director, a cameraman, a
sound recordist and an electrician (for anything more than minimal
lighting), plus as many assistants in each of these departments as is
necessary for efficient operation. There are few occasions when a
sound recordist actually needs an assistant for simple television
location work, except as a mobile microphone stand for certain types
of dramatic subject. For more complex set-ups, including many of
those for feature film shoots, a larger sound team may be required to
help with rigging microphones, cables and radio links, including the
link between sound control and camera.
Although radio links are now generally reliable, cable links (send
and return) between the sound equipment and the camera are, in
principle, safer. One potentially awkward decision is whether to
lay cables early, in which case on a big shoot others may lay their
own cables and equipment over the top, making changes difficult,
or whether to leave it late in order to secure easier access, but at
the risk of running short of time if there are unexpected demands
on sound.
A feature film may have a whole `video village' generating electronic and mechanical noise (e.g. from cooling fans on equipment),
304
Film and video sound
as well as operational noise: plainly, this will not be a good place for
sound, but the sound department must collaborate with it.
Camera noise has long been a problem for sound ± from the clatter
of the film camera onward. Noise from the camera has been reduced
by draping it with blankets (which is obviously an encumbrance for
the operator) or by wrapping it in a sound-absorbent `barney'. The
eventual near-solution came with better design: a self-blimped camera. It might be assumed that the switch to video cameras would
remove the problem entirely. Larger cameras (of which some of the
worst culprits have been in the first generation of HD cameras) may
require cooling fans. These can create a high-pitched squeal, perhaps
at about 12 kHz, well within normal hearing. If muted by draping it
with a coat, this may reduce it briefly, but then as the camera heats
up again, a second fan cuts in, shifting the problem to post-production ± where a notch filter will be required (but getting rid of the
noise is still not easy). Again, the only good solution is better design
in later-generation cameras: some already do not have fans. But HD
monitors, too, may have noisy fans.
A number of things will (or should) have been settled in advance;
these include the standards for the distribution of digital sound and
the camera frame rate, which will determine the time code generated
in the camera, and how that is distributed (possibly by radio link).
The frame rate for the edit and the rate for showing the finished work
may be different again. With good planning, well-matched equipment will keep problems to a minimum, but failing that can normally
be sorted out later: some time code gymnastics may be necessary
anyway. There are potential pitfalls: if picture is transferred to
NTSC for viewing, a delay line may be required to resynchronize
sound. On large feature film productions there may be a specialist in
charge of the entire chain: this certainly helps.
Film and video recording routine
Consider first the case where a film director has control over the
action before the camera.
If there is any rehearsal, this allows the recordist to obtain a microphone balance and set the recording level. In a full rehearsal the
action, camera movement and lighting are close to those to be
adopted on the actual take, so the recordist has a final chance to
see how close the microphone (or microphones) can be placed
without appearing in vision or casting a shadow, and can also judge
the ambient and camera noise levels in realistic conditions.
When the rehearsal is complete and performers are in place, a film
camera is made ready for identification of the shot, and the lights (if
used) are switched to full power. The recordist listens for and warns
the director about any rise in noise level from traffic or aircraft, or
unwanted background chatter. The director has checked with each
Film and video sound
305
department ± often very informally ± then (in the conventional start
to a scene shot on film) the camera assistant holds a clapperboard up
to the camera, and the recordist or sound assistant points a microphone to it.
On the director's instruction `turn over', the sound recording tape (or
other medium) is run: note that on digital audio tape a few seconds
run-up is required before the first usable sound in order to lay down
time code. Otherwise (for analogue sound) tape and film are run
together, and the recordist and camera operator check as their
equipment runs up to speed and the recordist checks for sync. With
`speed' confirmed, the camera assistant is told to `mark it'. Note that
although time code should be clear enough, a traditional clapperboard is still preferred by many filmmakers, as it shows unambiguous shot and take numbers and sync sound confirmation. It is
always used when the picture is on film.
[19.6] Synchronization of separately
recorded sound
A, Clapperboard showing scene and
take number (and other information).
B, Simple `blip sync'. 1, A light flashes
to blank picture. 2, Matching blip,
recorded on sound. This lacks the
precision of identification given by a
clapperboard, but offers greater
operational flexibility, and less film
stock is used.
Marking a shot (boarding or slating it) consists of stating clearly the
shot and take numbers, e.g. `one-three-seven take one', while holding the board up to the camera with the bar open. The board is then
clapped, the assistant and the camera operator both taking care that
the action of the board closing stays clearly in vision, and the
recordist noting that the clap is identifiable and distinct. If any fault
is observed, a second board can be called for and the camera
assistant puts the board back in shot, saying something like: `onethree-seven take one, sync to second clap', then clears the shot
quickly but quietly, while the cameraman checks the framing and
(if necessary) focus, and the recordist rechecks the microphone
position and gives a final thought to noise problems ± as the director
may also do at this time.
The director checks that the performers are absolutely ready, waits
for a signal that the camera is running at its proper speed, notes that
there are no new shadows or unwanted sound, then calls `action'. At
the end of the take the director calls `cut': the camera stops running
first, and as soon as it has, the recordist stops too ± or, if there is
some continuing noise, may signal for everyone to hold positions
while extra sound is recorded for a fade or mix.
Mute (US: MOS) shots, i.e. picture filmed without sound, are marked
with an open board (perhaps with fingers blocking any clap).
Wildtracks are recordings made without a picture. Identify them by
relating them to a particular shot number, e.g. `Wildtrack, extra
background of car interior with gear change for shot one-two-eight'
or `Wildtrack commentary for one-three-two onwards'. Alternatively,
use letters ± e.g. `Wildtrack B: atmosphere quiet schoolroom'.
Atmosphere or `buzz' tracks: the recordist should make a separate
recording of the characteristic continuing sound of each different
location ± this may be called `atmos' or `room tone' ± unless
specifically told it will not be wanted, and may also wish to
re-record important effects that can be improved by a balance in
the absence of the camera, together with any additional noises the
director wants.
306
Film and video sound
In the case where the director does not have complete control over
the action, it may still be predictable to the degree that a front board
is possible, or failing that, an end-board, which is inserted and
identified with the words `board on end'. With film, the recordist
should call for the camera to keep running if the director calls `cut'
before the shot has been slated. The director may instruct the camera
operator to take cues from the action: in this case the recordist
follows their lead. If a film synchronizing mark is missed, it is often
simplest to run the shot again and board it properly, rather than leave
the editor to find it.
Where a scene to be filmed would be disturbed by the usual procedure for marking shots, it can be simplified to the recordist speaking
the shot number and tapping a microphone in vision. For this, keep
the numbering simple, with no `take two'. A handclap, pencil tap on
a table or any similar action can also be used for rapid synchronization. When informal action is covered by a video camera with time
code, to be followed by digital editing, conventional shot-numbering
may be omitted but will generally be replaced by a comprehensive,
time-coded log, made (or checked) by using viewing tapes, if
possible before editing begins.
Film and video documentation
The director's assistant, the camera assistant and the recordist each
have their own documentation to complete. The production team can
work with domestic-quality viewing copies that have BITC (time
code in vision); these can also be used to make time-coded transcripts of dialogue. The director's shot list should include the
following:
.
.
.
.
.
.
Videocassette number or film roll.
For video, detailed time codes. For film, the board and take
number of each shot (perhaps adding some indication such as a
suffix `s' or `m' to the shot number for sync or mute; also `eb' for
end-board).
Shot description ± opening frame (on start of action); action
within the shot (including camera action; pans, tracks or zooms,
timed if possible); continuity points; and faults, errors or sound
problems.
The overall duration of significant action.
For film, the duration or footage used for each take (optional) and
`P' (`print'), `OK' or a tick, or `NG' for a shot not worth printing.
(When 16 mm film is printed as low-cost cutting copy it is
usually printed anyway, but the `P', `OK' or `NG' serves as an
editing guide.)
Wildtrack details (in the shot list between the sync takes at the
places where the sound itself is to be found).
The sound recordist's report should include as many technical
details as may be useful later, including camera speed, sound format
Film and video sound
307
and standards adopted. Analogue systems (if still used) may have
been controlled by time code or FM pulse, while two-channel DAT
has a single worldwide standard. Each shot for which sound is
recorded is listed, followed by a list of takes. `Good' takes are
marked: one way is to put rings round those numbers only. Wildtracks are included on the report in the order in which they were
recorded, and notes on quality may be added (see also notes on
documentation in Chapter 20).
In addition, the recordist identifies each new roll of (separately
recorded) tape over the microphone, with a standard list of details
that might include the recordist's name, name of production, any job
number or costing code, date, location, recorder type and number,
and (in case of doubt) camera speed. This may be followed by a
10-second burst of reference tone at an agreed (or stated) standard
volume such as `minus 8 dB' (i.e. 8 dB below nominal maximum).
Unless otherwise agreed, the film sound recordist is responsible for
getting tapes back for transfer within the shortest convenient time:
the simplest arrangement is for them to accompany videotapes (or
film, which if possible is sent for overnight processing) and then
forwarded with the rushes for transfer while the picture is viewed
with any guide audio from the camera (or while film rushes are
viewed without sound). This assumes a fast, routine operation, but
even in a slower schedule, film would still be viewed for technical
faults as soon as possible, though sound might not be synchronized
with it until later.
The first, preparatory stage of editing is to identify all raw materials
clearly, and then to review and `log' them. `Prepping' includes
checking through content in enough detail to know (with the help
of timed notes) where any major component is to be found. The log
may include timed `in' and `out' cues, as well as highlight points and
notes on participants and content. Ease of editing will often depend
on how quickly obscure material can be located ± often things such
as a single spoken word with a particular intonation that would not
usually be included in a log.
If time and budget allow, a full transcript of unscripted speech can
save a lot of time. The transcription typist should be told in advance
who the voices are (and how to identify them), be provided with
dates, times, places, subject matter and unfamiliar spellings; and also
advised on how much to include from off-microphone questions,
director's instructions, noises off, background atmosphere, interruptions, and errors and idiosyncrasies of delivery. Transcripts should
be supplied on disk as well as on paper. Users often have their own
preferred layouts and may want to be able to select material from it,
to assemble a script of work in progress or the final edited product.
A transcript can be marked up by eye, rapidly skimming through it
prior to logging, then checked out in more detail against recordings.
Individual words and phrases, too, can often be found by eye. Word
processor `find' searches can be used for this, but are fallible unless
a transcript has been checked for errors that might not show up on
common spell-check programs.
308
Film and video sound
Videotape editing
A variety of techniques has been used for editing picture and sound
recorded on videotape. One major division is between off-line and
on-line techniques. Where editing is done off-line the on-line process that follows it is essentially a computer-controlled copy-editing
system. Editing is further divided into linear and non-linear techniques. The term linear means that everything downstream of a cut
has to be recopied, as in any copy-editing system. In non-linear
editing a change can be made in the middle of a section without
affecting material on either side. It is non-destructive. Film editing is
a non-linear process.
On-line linear editing may begin in a two-machine suite, progressing
to three or more replay machines at a final (and more expensive)
on-line picture editing session. But in any linear editing system
flexibility is severely limited, either to few editing stages or in
complexity. Film, in contrast, becomes easier to recut in the later
stages, so the process of refinement is not halted by diminishing
returns for the effort and time expended. Film editing was an early
example of non-linear editing.
Given its greater flexibility, film editing was unlikely to be fully
superseded by videotape editing. However, digital editing systems
are non-linear, just as film is, but much faster, which makes it far
easier to try out different ideas for both picture and sound, and also
speeds consultation between editor and director. As a result, digital
off-line video editing is increasingly used in the film industry as well
as in television. If (in many cases) it now takes as long as film
editing once did, this is because it allows more opportunities to work
on finer detail and complex effects. As with so many advances in
digital technology, its main benefits have not been in cost or time
reduction, but in permitting more ambitious and sophisticated
results, while maintaining technical quality.
Digital video editing
Professional digital video editing requires a substantial investment
in dedicated hardware and software ± carrying the associated bugs
and glitches that inevitably seem to emerge in such a rapidly developing field in which frequent upgrades are offered. Initially, the
most severe limitation was in memory for picture, but in off-line
systems about 10 GB of memory offers adequate (reduced) picture
quality for editing most television productions of an hour or so if
material is loaded selectively, while 60 GB gives much greater
flexibility, and access to a server increases this further.
As computer processing power and memory (with recordable data
tape stream as backup) have dramatically increased, off-line non-
Film and video sound
309
linear editing can also, if required, be replaced by on-line non-linear.
This has extended the duration, from initially just component items,
up to complete productions for which no picture compression is
necessary. It also means that relatively simple editing (to a standard
that just a few years earlier would have been regarded as highly
sophisticated) can be achieved even on good domestic equipment by
using editing software that costs very little, compared to the top
professional systems.
The director's time-coded shot list is used first to make a preliminary
broad selection from the material available. From the shot list (and
transcripts, if available) the director can search out the preferred
scenes, takes, segments of interview or other selections. The in and
out time codes are entered into a program that can be used to copy
from the original tapes directly into the memory of the digital editing
system without the intervention of the editor other than to feed in the
tapes on demand. As the selections are loaded, this is also a good
opportunity for the director and editor to view the material together
and discuss their objectives. If there has been no time for preparation, the time codes may be entered manually as the rushes are
reviewed.
Audio tracks are loaded in a digital format. If the sound is sampled at
44.1 kHz (as used on CD) or 48 kHz (the standard for DAT) it will
remain at broadcast quality, and can be treated as on-line, i.e. the
product of the edit can be used in the final version of the sound.
Lower sampling rates are acceptable for editing where sound as well
as picture will be recopied (conformed) later to match the edit
decision list (EDL). On-line-quality sound can be cut alongside
either edit-quality or on-line-quality picture.
Picture and sound are loaded together, if this is how the material was
shot, or separately in the case of mute picture or wild sound. It is
helpful if the system also records and holds a few seconds more
sound at either end of the selected picture in and out points: this will
allow for mixes or `soft' cuts without having to take extra care,
except in cases where there is unwanted sound very close to a cut
that will become a mix. During replay, a marker travels along timeline view-strips that look (for those with experience of film editing)
much like film and audio tracks laid out across the screen one above
the other. These strips are also used to show cuts and fades. A simple
display might have just the picture and one (mono) or two (stereo A
and B) tracks, but with provision for many more. In principle it is
possible to have an unlimited number of tracks ± although with a
visual limit to the number that can be seen on the screen at any one
time.
To make an edit, a shot (or some other length of picture or sound) is
marked, then some chosen operation carried out. In this way, complex edits can be undertaken, and if they do not work little time is
lost, as almost any change can be reversed by an `undo' command,
and the choice of system will dictate how many stages it will be
possible to go back in this way. (But if a long sequence of operations
is envisaged, the edits up to the starting point should be saved, so
310
Film and video sound
they can be reloaded, if necessary.) This, together with the usual
computer drag and drop and cut and paste facilities, makes digital
editing both easier and much faster than the conventional film
editing it is otherwise very similar to.
One of the few disadvantages of digital over film editing is that
computer programs can crash, so it is essential to back up regularly
to ensure that a recent edit and perhaps also some earlier versions are
always available. Another is that the increased speed reduces thinking time. A documentary, for example, may require commentary,
and if this is written in the editing suite it may slow down other
work, so it can sometimes be helpful to divide the editing with a gap
of, say, a week or more in the middle ± although this may be
impracticable for scheduling reasons.
Editing sound to picture is more complicated than sound on its own
(as described in Chapter 18) in that, as the picture is opened out or
condensed, all the tracks of sound must have a corresponding length
(an integral number of frames) inserted or deleted. Shots or
sequences can be inserted, pushing the existing sound or picture apart,
or replaced, overlaying it. Keeping it all in sync requires care.
[19.7] Avid hard disk editing
1, Keyboard and mouse to control a
range of display windows on two
screens. 2, Picture monitor. 3, Edit
monitor, here operating in `trim'
mode, showing pictures before and
after a cut, with a bin of clips and the
picture and sound timeline below.
Digital editing systems must operate in a range of modes, which
include initial loading (which for film or analogue sources includes
digitizing), then editing and, after that, tracklaying. It is best to do
most of the editing with just one track or a stereo pair, at first mostly
with parallel cuts, then at a later stage putting in simple overlays.
When the picture edit is virtually complete, tracklaying begins. Even
then it is still possible to recut the pictures, although more laboriously and with an increased possibility of confusion over sync.
[19.8] Avid `manual user interface'
This is one of a range of control
options.
One leading system (Avid) has two screens, one showing pictures
where the effects of small changes can be seen, the other offering
information (timeline, access menus, bins of clips, etc.) and some of
the control icons for editing. As for the audio editing described in
Chapter 18, the sound waveform is shown. By using variable-speed
replay the sound can be finely matched to the waveform, and a
cutting point selected. As usual, for a clean cut, avoid cutting where
the waveform is dense: a gap is better. Again, there is a `zoom' that
condenses or expands the timeline. Zooming in makes detail of the
waveform clearer, so that a finer editing point can be located.
Zooming out reduces detail but puts more (eventually all) of the
production on screen, which is useful for finding and moving to a
different place.
[19.9] Avid sound editing and
tracklaying displays
Picture and tracks are lined up in
sync, as for film. The audio
waveforms help the editor to find
cutting points to within a small
fraction of a frame.
An alternative system (Lightworks) employs a single, large screen to
display pictures, tracks and a wide range of filing systems, each with
a separate window that stands out like a three-dimensional block
seen in perspective, with control icons that may be ranged along the
top and down one side. There is also a keyboard with function
buttons and an external VDU that displays data. It is useful to add
an external monitor for the output picture and a conventional PPM.
The main screen window areas include one or more `viewers' for
displaying selected individual pictures, `racks', or filing cabinets of
Film and video sound
311
time-coded and annotated lists of shots and `galleries' of scenes
(banks of tiny initial frames with their designated titles below them).
For sound, the most important window is the strip-view with a scale
of seconds along it. In its simplest form the strip corresponds to 10
seconds or so of picture with the sound below it and, as usual, it is
possible to zoom out for a broader view or in for detail. Cuts are
indicated by changes of colour and cues by displacement of the strip.
From the start, Lightworks mimicked film transport by including a
large flatbed-style lever control to run picture and sound at variable
speed back or forward in sync, in order to be able to stop precisely at
the point required for a perfect sound cut, even though this might not
be on a junction between two frames. Gaps, sound effects or changes
in background or voice quality can be quickly located by listening
for them, as with film. This kind of facility has been widely adopted.
[19.10] Lightworks editing system
1, Editing screen, with layout chosen
by the editor, here including two
`viewers' (windows displaying
pictures running in sync); a `rack'
containing the material and
information on selected groups of
shots; a `gallery', here of eight shots;
and a view-strip with a segment of
the timeline of picture and two audio
tracks. 2, Keyboard. 3, Mouse.
4, Flatbed variable-speed control. At low
speed the sound `scrub' is heard at
proportionally reduced pitch, as with
film or audiotape, helping the editor to
locate sound cutting points precisely.
Plainly, as sound and picture editing progresses, it is vital to maintain synchronization of all tracks downstream of the point on an
individual track or stereo pair where changes are being made ±
which means that an editor must also be skilled in recovering sync
if by mischance this is lost. Further, any technical problems encountered when learning such new techniques can be frustrating and also
expensive in studio time, if this is locked into the use of the digital
equipment.
For sound with picture, as for sound alone, the editing process ± the
stages and objectives at each stage ± are much the same. Except for
the simplest edits, it begins with a rough assembly, followed by
more detailed editing of individual sequences.
Rough and fine cutting
The editor usually works with the picture and as few sound tracks as
possible until the picture cut is complete. The first stage of editing is
to make a rough cut, almost invariably at this stage cutting sound
and picture parallel, making `straight' or `editorial' cuts. Where
picture and its related sound have been shot (or recorded) separately,
and so may differ in length, matching is sometimes just a matter of
making the longer one equal to the shorter with a parallel cut. Where
picture is longer than the sound intended to cover it, the picture can
usually be allowed to run its proper length, with a gap in the sound
that will later be filled out with matching atmosphere or effects.
[19.11] Lightworks view-strip
V, Segments of picture, each
represented by a different block of
colour. A1 and A2, The associated
sound, here as two tracks with, right,
faders that can be clicked and
dragged to new settings. The upper
and lower parts of the fader `knob'
can be moved separately to set
volumes before and after a fade, the
ends of which are marked on the
associated strip.
Cuts may be made by removing corresponding lengths of sound and
picture. A short cut in picture can often be accommodated by making
312
Film and video sound
several shorter cuts in sound that add up to the same duration,
perhaps by taking out hesitations or inessential words. Opening out
sound, and in particular commentary that has been recorded wild
(i.e. not to replay of picture), in order to match picture is a common
and often desirable device, because it is better to have more picture
than words. Lengthening picture to match sound is an exceptional
measure ± and requires a better reason than just to solve an editing
problem. In all cases, the timing of speech, both in its spoken rhythm
and in its cueing, should remain comfortable and natural. One of the
objectives of the fine cut will be to restore these qualities if they
have been lost or diminished in the rough assembly.
At the fine cut, as further sound and picture are trimmed away,
sound overlays (where the sound from one shot runs over the next)
may be introduced. Before any complex non-parallel cut is made,
the editor ensures that sync can easily be found again both before
and after the region of the cut. Much of the technique for the detailed
editing of sound on video and film is identical to that described
earlier for audio tape: the procedures for choosing the place are just
the same, but there are a few additional points to note:
.
.
.
[19.12] Lightworks display elements
1, Viewer: a simulated 3-D box with
the picture on the front and control
icons and data displays on two sides
and plinth. 2, A gallery, in this case of
eight selected shots, with the names
or numbers that have been allocated
to them and icons that allow the
editor to identify, select and use
them. 3, Part of a recording panel,
with controls and a file card.
Words, syllables or pauses can often be lifted from rejected takes;
even if it results momentarily in a slightly awkward mouth movement, this may still be preferable to inappropriate words.
A change of sound that happens with a cut is much more natural
and acceptable than an unexplained change at another point.
Short gaps in background atmosphere created by lengthening
picture (e.g. with a cutaway that needs to be seen for longer than
the master action it replaces) can sometimes be filled with closely
matching or looped sound. This is better than trying to find
similar effects from library sources, as it takes a lot to obliterate
a drop-out of atmosphere, particularly where there is a continuous
steady noise such as machinery hum or the whine of jet aircraft.
In professional work, there is often a stage at which the work is
reviewed by someone with executive responsibility for the result
(and the power to require changes). It may be sensible to leave a
final stage of fine cutting until all major editorial changes have been
made. These may also affect overall timing.
Shot list, commentary and music
A timed shot list of the edit (often read as progressive time code
from the first frame) may be required for a variety of reasons: for
documentaries these include commentary writing. Narration will
often have been drafted earlier and revised as editing progressed,
but it is better to complete the edit, so that the exact length and value
of each picture is known, before the final version of the words is set.
(For film, time may be replaced by `footage', where each foot is
taken to be 16 frames, as it would be on 35 mm film: these are a finer
measure than the 40-frame 16 mm alternative.)
Film and video sound
313
The shot list can conveniently be combined with a script of the
dialogue as cut and notes of other important sound. This can be laid
out to look like the standard television script with picture on the left
of the page and the corresponding sound on the right. Times should
be noted not only at cuts in picture, but also at important points
within the action of a shot (e.g. when some feature first appears in
vision) and also at the start and end of dialogue, strong effects or
dominant music.
Commentary can then be written accurately to picture: for a fastspeaking narrator allow a maximum of about three words per second
(on film, two per 35-mm foot). For slower speech or more expression,
these figures must be reduced. It is far better to write too little than too
much; in any case, strong pictures are often better without words.
The audience's primary need from a commentary is that it should
interpret the picture: this is a more effective use of the medium than
when the pictures are cut like wallpaper to cover the words (`wallpaper' is a term of derision). The worst possible case is when
pictures and words do not agree, either because the commentator
wants to talk about something else or to go on about something that
is no longer in the picture. In these cases an even moderately
interesting picture demolishes the commentary entirely: the words
are not heard.
There is a particular danger when a presenter records part of a
commentary in vision, either in the studio or on location, and another
part later in the confined space of a commentary booth. Even if the
voice or microphone quality can be matched by equalization, there
may be an abrupt change of acoustic. Short links should, if possible,
be planned in advance and recorded at the location; for safety, a
variety of versions of the words might be recorded.
At best, a succession of different acoustics on the same voice, for no
apparent reason, is mildly distracting. At worst it is destructive of
artistic unity and also confusing: the audience may be unsure
whether a new sound is the same voice or some new and unexplained character. On the other hand, well-chosen words, written to
fit and complement the pictures, may add far more than is lost by the
mismatch in sound. If so, one compromise is to allow a gap in time,
a music link or strong sound effect to intervene: the change will then
be less noticeable.
If it is obvious from the picture that the voice-over was not filmed on
location, the change in quality may be helpful: the commentator
finishes speaking in vision, turns and walks into the action, and the
dubbed voice takes over: here a clear convention has been adopted,
and the change of acoustic confirms it.
Specially composed film, television and video music is often written
at a late stage. The composer requires a copy of edited picture and
sound that is as complete as can be arranged, a cue list with time
codes or footages and also a script (for documentary, a shot list with
narration that is as advanced as possible). However, after the music
has been composed and recorded, the editor will often find that the
314
Film and video sound
interaction of picture, words and music suggests new possibilities: it
is wise to allow editing time to allow for this. Final changes in the
picture may in turn require changes (further cuts or extensions) to
the music, so because it is often written and recorded to a tight
deadline, it is worth making sure that there will be sufficient
flexibility.
Music from existing sources should be selected earlier in the editing,
so there is more time for the pictures to be accommodated to it.
Tracklaying
Tracklaying is the first stage of preparation for dubbing (making the
final sound mix). Here again the standard techniques were pioneered
with and are still used for film, and in digital editing are functionally
almost identical. In all media, when the fine cut is complete, the
constituent parts of the cut sound are separated on to several tracks:
this is called `splitting the tracks' (on film, a physical process).
For as long as the sound continues at the same level and quality,
whether of the main or background sound, it is kept on the same
track, but at points where these change, the sound is switched to
another track alongside it. If, by simple adjustments of level and
equalizer settings, it is possible for the sounds to be matched, a
parallel shift from one track to the next may be all that is necessary.
But if there is a more drastic change that is not immediately
explained by an equally strong change of picture, then extra sound is
required for a cross-fade. These fades are generally made quickly,
but it is better to have too much than too little, or the sound may be
heard cutting in or out at low level. The chopped-off sound appears
unnatural if it is not at a cut.
For film, extensions had to be found from a trim bin. On digital
editing systems they should be easier to find: in some, overlaps
reappear automatically as the sound is split onto neighbouring
tracks.
During the early stages of editing, extra sound that had to be
synchronized with action may not yet have been cut in, perhaps
because there was enough other synchronous material for viewing
purposes. At the tracklaying stage extra tracks are created for them,
and also for any commentary that has been pre-recorded.
Music may also be laid: where it is dominant it will already have
been arranged for the picture cuts to be related to it, or alternatively
for specially composed music to match the picture. The precise manner
in which picture and music should be co-ordinated is a complex study,
but a strong rhythm suggests that some, but not too many, of the cuts
are related to the beat; similarly, strong notes may cue a cut.
The editor accurately lays as many of the sound effects as possible;
this includes both library effects, which should be selected and
transferred for the purpose, and wildtracks that were specially shot.
Film and video sound
315
Some continuing sounds may be formed as loops. The name reflects
the way this was once done: by making up lengths of recorded track
with the ends joined back to the beginning. Now this is replaced by a
digital `loop', using material selected and repeated from a sampler or
disc. Loops with strong distinguishing features must be long, so that
the regular appearance of that feature does not become apparent. If it
is likely to, it could perhaps be trimmed out, though this would shorten
the loop even more. In joining the ends of loops (i.e. by selecting the in
and out points of a sample) take care to find points where volume and
quality are matched, or where overlaps can be mixed.
The documentation required to go to a dub varies both with media
and with the complexity of the operations that will follow. With film
(or in any case where the tracks do not appear on screen) the editor
or an assistant prepares a chart showing the main points of the
action, and parallel to this the sound on each track, or source,
showing the cuts in and out of sound with symbolic indications of
fades and mixes and clearly marked time codes or footages, together
with notes on loops and tape and disc or hard disk effects. If the
tracks do appear on a screen, this is (in principle) simpler ± although
aids (such as cues and colour codes) may be added, to make them
easier to read. A marked-up script is helpful ± or essential, where
commentary is to be added.
Conforming sound and picture
When an off-line or digital (or videotape) edit is complete, the
picture must be conformed (matched and copied) from the originals:
this is equivalent to the negative-cut and showprinting of film.
When the off-line picture has been accompanied by on-line quality
of sound, this must be transferred back in sync with the picture. In
the more general case, the sound as well as picture will be conformed,
which means that it must be separately recopied from the original
recordings. Tracks which have been laid off-line as a provisional
set of editing decisions are used to create a corresponding set of
full-quality audio tracks in a format that will serve as sources to be
used in the mix. For this, there are two main options. The older
method is to copy the originals to multitrack tape; the alternative is
to use hard disk. An edit decision list (EDL) for picture and audio is
used to control the conforming process automatically.
To do this, the EDL finds and displays the first time code, including
the user bits that identify the source: in effect, it `asks for' a recording. When this has been inserted, it will then cue and transfer all the
material on it, leaving gaps for material that will be inserted from other
recordings; then it will request a second source and so on. Picture is
generally conformed to a single, continuous recording, and dissolves
are programmed or put in manually later. In contrast, sound is transferred with the tracks alternated in a checkerboard A- and B-roll
arrangement, so that they can be separately mixed and controlled.
316
Film and video sound
Where television to be transmitted with stereo sound has been edited
using equipment on which the editor, for simplicity and to minimize
the danger of losing sync, has been working in mono, a clearer idea
of what is available in stereo or surround sound may emerge only at
this stage. Any stereo in the MS format is converted to A and B.
Note that for the benefit of television viewers who do not have stereo
decoders, television sound is still also transmitted in a form in which
the A and B signals have been recombined to form a `compatible'
mono signal. Where encoded (and so compressed) surround sound is
transmitted, in order for it to be heard, this will require decoding at
the receiver; without that, normal stereo will still be available.
Stereo and surround should not be made so extreme that mono
listeners cannot hear all the information they need to make sense
of and enjoy the production.
Complex sound preparation is now often undertaken using a digital
audio editing program, either a separate stage as described in
Chapter 18 or (on a digital console) as an extended first stage of
the process of mixing. Even if the previous tracklaying has been
comprehensive, preparation for the mix may be a time-consuming
process that requires equipment to replay audio from a range of
different formats. From being one of the more neglected components
of video editing, `prep' has developed to the stage where it may
occupy a room in its own right, in effect a small studio. Depending
on the facilities available, some tracklaying may be left until this
stage; some of the sounds may be processed here, automatic dialogue
replacement (ADR) and spot effects (`foley') laid and edited, and
some sequences pre-mixed. In such a case, the preparation may be
undertaken by a specialist sound editor who carries this to a stage
where the final mix is much simplified ± except that the theatre also
serves as a meeting place where production can hear what is offered
and either accept or change it on the spot.
Post-synchronization
Post-synchronization or automatic dialogue replacement (ADR; not
necessarily very `automated' by current standards) may be needed in
order to re-record voices that were inadequately covered or overlaid
by inappropriate background noises, or because a director has given
priority to a costly setting or unrepeatable event. It may employ an
improved version of the original speaker's voice or substitute a more
appropriate accent, vocal quality or performance. It may replace one
language by another. ADR is a specialized job that must be done
well if it is to be effective.
Good post-synchronization has to be done phrase by phrase using
words that give reasonable approximations to the original mouth
movements. A loop of film or its digital equivalent is repeated over
until the best match is obtained, then the next phrase is taken with
the next loop. Any final trimming or tracklaying to match lip move-
Film and video sound
317
ments perfectly can generally be done afterwards. Near-perfect
results can sometimes be achieved by digital editing.
Post-synchronization of spot effects (`foley') is ± for a few talented
individuals ± a well-paid job in the film industry. These experts are
capable of seeing through a scene or sequence a couple of times,
then collecting together a pile of odds and ends which they manipulate in time to movements both in and out of picture, in a convincing simulation of the casual sounds of live action. In a low-budget
production this may be sufficient, but in the more general case, the
sounds will be refined by further editing, tracklaying and treatment.
A complementary technique (at a much earlier stage) is shooting
picture to playback of pre-recorded sound ± typically for musical
production numbers (miming), where the quality and integrity of the
sound are of primary importance. It also permits live action (such as
dance) to be combined with song in a way that would be difficult if
not impossible in real life. In miming, however, the illusion can be
shattered by a sudden change of acoustic between speech and song,
emphasizing and making awkward a transition that should be smooth.
Mixing sound for picture
To mix sound for picture, the theatre (dubbing theatre) has equipment to display picture and run many synchronized audio tracks, and
then to record it. There is often also a studio, if only for a single
commentary voice, and provision for playing CDs, tapes, digital
audio files, etc., together with all the additional facilities on the
control console that can make operations simpler and more flexible.
Since digital processes have entered the dubbing theatre they have
allowed video and film dubbing to converge and become independent of both the original recording medium and its final format (film,
[19.13] Film re-recording studio
(dubbing theatre)
1, Mixer at control console.
2, Assistant, with tape, cassette
decks, turntables, etc. 3, Film
projector and footage counter (here
with video cameras to relay picture
and footage as required). 4, Magnetic
film sound reproducers and recorder.
5, Loudspeakers. 6, Screen and
footage monitor. 7, Narrator with
desk, microphone, cue light and
monitors for picture and footage.
8, Acoustic treatment.
318
Film and video sound
DVD, videotape or television transmission). These are some of the
components that may be required:
.
.
.
.
.
.
.
.
.
Picture display ± in digital systems, a simple monitor, plus
another for any narrator or commentator if the main screen is
out of sight.
A counter showing timings or time codes in minutes and seconds
should be displayed close to or superimposed on picture, perhaps
with repeaters on the console and in the studio. A narrator or
musical director works to these numbers together with other cues.
Replay and recording equipment. This may be under local control
while being loaded, but then switched to remote operation so that
synchronous replay of all sources and recording machines is
controlled from the console.
Ancillary replay equipment. There must be the normal provision
for replay of various recorded formats, for original recordings,
music and effects. There should also be provision for the replay
of sampled sounds as loops.
Commentary booth, usually with a dead acoustic. This ensures
that narration does not carry an ambience that is alien to the
picture it accompanies; for example, by overlaying a live, indoor
acoustic on to an exterior scene. A dry acoustic also seems to help
the voice to cut through background effects or music.
Cue keys and lights (for a narrator to be cued from the console).
The usual system is `green for go'; the cue light should be close
to the reader's sight line. The narrator will often require a
headphone feed of music, dialogue and effects, with or without
voice feedback (whichever is preferred). Before rehearsal begins, check that the sound level fed to the studio headphones is
comfortable. There is also talkback between control console and
studio.
Provision for the production of spot effects (`foley') with a range
of floor surfaces, perhaps in the same acoustically separate booth.
Monitoring equipment ± high quality, and usually set loud
enough for imperfections in the original material or of the mix to
be heard easily. Also, as usual, there should be provision for
the sound to be switched to domestic quality loudspeakers to check
what most of the audience will hear ± particularly important
when mixing dialogue or commentary with music or effects for
television.
The control console (see Chapter 9). This has the usual fader
channels, each with its standard facilities for equalization and
feeds to graphic filters, artificial reverberation, etc. Compressors,
limiters and expanders are available, and also a noise gate. There
will be a full range of controls for stereo and generally also 5.1
surround, including panpots and spreaders.
There is provision for discontinuous recording, sometimes called
`rock and roll'. The mix continues until errors are made, when the
sound and pictures are stopped and run backward until well before
the point at which re-recording must restart. A counter runs back in
sync with sound and picture. Then the whole system is run forward
once again, and the mixer checks levels by switching the monitor
rapidly between the input and the previously recorded signals,
Film and video sound
319
[19.14] Film dubbing cue sheet
The original dub on which this was
based has provision for more sound
rolls. Here 35 mm film footages are
given (i.e. 16 frames ˆ 1 ft) but timings
may also be used. The action column
does not list every picture cut, just
sequences within which sound should
remain roughly the same. In some
cases sound has been laid to several
seconds beyond a picture change, but
will be faded out quickly on the cut
(this is represented by the shallow V).
As when dubbing in any format, fine
points are discussed or decided by
consultation on the spot. For example,
here it was not made clear on the
chart, but the choir was taken behind
the boys' chatter between 3 and 17
and was faded quickly after 17,
although it was originally laid as far as
32. The rugby football sound (at 16)
preceded the cut at 17 and, in effect,
killed the sound on tracks 2 and 3.
`V/O' means `voice-over'. At 80±84 the
teacher's voice was to be heard over
background chatter and effects, and
from 84±105; this in turn was taken to
low level while Richard's voice was
heard (out of vision) talking about what
is happening at this point in the story.
`Hum' and `buzz' are terms used to
indicate low-level background sound.
The commentary voice had been
recorded in advance and laid
(on track 4).
adjusting faders to match them, then switches back to record, and
from that point the new mix replaces the old. The switch should not
be instantaneous, of course, or a click might be introduced.
Documentaries are often mixed in stages, with a so-called music and
effects (M and E) mix being made first (in fact, this would also
include any sync speech, and there might be no music). The final
stage is to add the commentary; but the M and E track should still be
retained against the possibility of changing the commentary later, or
making versions in other languages, or for use when extracts are
taken for other productions. In fact, on some desks, a mix at any
stage can be held in memory and subsequently read out automatically if needed (for example) to fulfil contracted specifications for
`deliverables'.
320
Film and video sound
Narration (commentary) and effects can be added at this stage.
`Green light' cues for speech have to be given a little early to
allow a performer time to respond; if desired, fine adjustments to
the timing can be made later by shifting the recording by a few
frames.
[19.15] Film: 16 mm separate
magnetic sound
1, Edge track used in US and Canada.
2, Centre track used in Europe. The
full width of the film is coated and can
therefore be used for either system.
Narration recording is often treated as a separate operation, for
which a timed script with marked cues is provided. This script can
leave out most of the picture details, and show only the most
significant details of other sounds, such as the in and out cues of
existing dialogue or featured sound effects. If narration is recorded
before the final mix it can be taken back to the cutting room to adjust
each cue precisely to the frame at which it has the best effect.
Sometimes, fine trims of picture may be removed or restored at this
stage: occasionally the strength of the word enhances a picture to an
unexpected degree and demands that it remain on the screen a little
longer. When this fine-tuning is complete, the tracks are all returned
to the dubbing theatre, and the final mix is completed.
Film sound editing
[19.16] Film: 16 mm combined
magnetic sound
The narrow magnetic stripe occupies
the space formerly used for optical
sound. Picture and sound are
displaced to allow for the two sets of
equipment required for synchronous
reproduction.
[19.17] Flatbed film viewer
Once very widely used for film
editing, this is a simple example of
the equipment for viewing 16 mm
film at normal speeds and with
moderate picture and sound quality.
1 and 2, Picture feed and take-up
plates. 3 and 4, Sound feed and takeup plates. Lacing is simple, with drive
spindles on either side of the central
sound and picture heads. It is not
necessary to have rolls of film as
shown here: the machine requires
only enough head or tail of film to be
held by the drive rollers. Typically, the
film can be run at normal and `fast'
speeds, and either forward or
backward. For editing, an extra sound
path is needed, so most flatbeds
have six plates. Further variants allow
for stereo sound and other features.
This chapter ends with a description of some of the older techniques
used for film sound (and picture) editing. This begins with transfer.
When separately recorded audiotape is transferred to separate
magnetic film, the recording speed is controlled by time code (or
sometimes a sync pulse) from the original. The original recording is
kept as a master, and can be used for retransfer of worn or damaged
sections, or for finding extra sound for transfer to digital sound
file or for direct, unsynchronized replay to fill gaps at a dubbing
session. It is simplest to transfer stereo ± MS or AB ± in the form
received, unless otherwise specified, but MS will be converted to
Film and video sound
321
[19.18] Editing bench
1, Picture synchronizer.
2, Amplifier and small loudspeaker.
3, Horse supporting rolls of film and
spacer. 4, Winding spindle with `split'
reels fitted. 5, Bins lined with soft
fabric, to hold excess film that has not
been wound up. 6, Racks for small
rolls of film awaiting selection of
material.
AB when sound is being prepared for the dubbing (re-recording)
session.
In the cutting room the film editor receives:
.
.
.
.
.
.
.
[19.19] Synchronizer
`Pic-sync' with simple loudspeaker.
1, Picture track. 2, Lamp projecting
picture via rotating prisms to the
screen (3). 4 and 5, Sound paths over
replay heads on centre line: each path
passes over a wheel with (for 16 mm
film) 40 sprockets, a `16-mm foot'.
6, The wheels are on an axle that is
turned by hand, but some models
also have a motor and clutch plate, so
edits can be checked at normal
speed, and a disc indicates the frame
within each foot. 7, Counter showing
feet or time from when it was set to
zero. 8, Simple low-level sound
mixer. 9, Loudspeaker input.
10, Power transformer. Synchronizers
may have two picture heads or more
sound paths and a more
sophisticated sound system.
film positive, `rush prints';
the original sound tapes;
sound transfers on the same gauge as the original film;
documentation on the film and sound, and perhaps also laboratory and viewing reports on the rushes;
a shot list;
a script, or sometimes a transcript of the sound;
alternatively or additionally, a cutting order, which may include
shot descriptions from the shot list and dialogue from the
transcript.
322
Film and video sound
The cutting room has a flatbed viewing and editing machine,
tape joiner, wax pencils, trim bin, leader film, spacer film and a
waste bin.
[19.20] Looping back
A mechanical technique for matching
sound to picture: these are in sync on
both sides of the `pic-sync', but at the
picture join there is an excess of
sound, which has been taken up in a
loop between two of the sound paths.
In this way, picture and sound can be
rolled back or forward together until
suitable points for cutting the sound
are found. Similarly, excess picture
may be removed to match sound.
There is also a film synchronizer (with its own small amplifier
and loudspeaker) on an assembly bench. This has two bins, lined
with soft fabric, into which picture and film sound can fall, and
between them a platform wide enough to hold the synchronizer.
On either side of the bench, outside the bins are supports for the
rolls of film and sound to allow them to spool across the bench.
On the left side is a `horse', several uprights with holes, so rolls
of film can be supported freely on spindles passed through the
plastic cores on which film is wound. On the right is a take-up
spindle to hold a number of split spools, with a winding handle to
take up film.
The synchronizer has sprocketed wheels, rigidly mounted on an axle
so they are mechanically synchronized. Each wheel is 1 ft in circumference, so that for each revolution 1 ft (16 frames of 35 mm or
40 frames of 16 mm) of film passes through the synchronizer.
A counter shows feet of film. Even for 16 mm film this is often
calibrated in `35-mm feet', so that 16 frames equals one digit on the
counter. There is also a disc that numbers each frame on the circumference of the wheels: on 16 mm equipment these run 1±40.
Picture is carried on the front film path and sometimes on the second
as well. The other paths have sound replay heads. The synchronizers
can be turned either by hand or by a motor that is engaged by means
of a small clutch so that picture and sound run at their normal speed
± forward or backward.
Film and sound are cut by stainless steel (and therefore nonmagnetic) blades on a tape joiner (guillotine splicer), which then
holds the ends in position while tape is stuck across the join. The
tape recommended for joining film is similar in appearance to
transparent plastic domestic tape but is polyester, very thin and
strong, with a non-oozing adhesive. Tape joins can be undone, to
remake edits or reconstitute the original without loss of frames. Wax
pencils are used to write on film `cel'.
[19.21] Tape joiner
1, Picture or magnetic track is placed
on the lower plate and located
precisely over the sprockets that
stand along one edge. 2, After knife
has cut film, it is moved to the centre
of the plate and the film to be joined
with it butted to it, again with
sprockets holding the two ends
precisely in position. 3, A strip of
adhesive tape is drawn across the join.
4, Tape also adheres to two bars, one
on each side of the base plate. 5, The
upper head is then brought down on
to the film. 6, When the spring-loaded
arm is pressed down, two guillotines
cut the tape on either side of the base
plate, and a punch perforates the
adhesive tape at the sprocket holes
near the centre of the plate.
The standard angle of cut provided on some joiners used for magnetic film sound is 10 . A mono recording on separate magnetic film
track is 5 mm (0.2 in) wide: a cut therefore fades in over about
0.5 mm (0.02 in); for 16 mm film the duration of the fade is 2.5 ms,
quite long enough to avoid a noticeable click. A right-angled cut is
also available.
A trim bin is used for hanging up sections of picture and sound that
have been cut. Sound trims may be needed later to overlap sound in
order to cross-fade between scenes.
Blank spacer film is inserted in the sound assembly where no sound
is available or in the picture where a shot is missing. Spacer is also
used to protect the ends of rolls of assembled film. Leader film for
picture and sound is printed with start marks and 35 mm footages
counting down from 12 to 3, after which the print is black to the
point at which the film starts (which is where zero would be on this
Film and video sound
323
[19.22] Synchronization marks
1, Marks on picture. 2, Marks on
separate magnetic sound. In the
conventional clapperboard system
these are marked at the points where
picture and sound close. When
electronic synchronization is used,
they mark the flash on the picture
and the corresponding buzz on
sound.
count). Sometimes a frame of tone (a `sync plop') at 1000 Hz (for
16 mm) or 2000 Hz (35 mm) is cut in on the sound leader a few
frames before the film starts. As it passes through prior to being
switched, the presence of the tone confirms that sound is live on the
replay itself.
The principles of cutting separate magnetic film sound are mostly
similar to those used for audiotape, though a few extra problems
(and possibilities) emerge from the combination of sound with
picture.
Film picture available to the editor includes:
.
.
.
film shot with synchronized sound ± lip-sync sound, synchronized effects or music, and material where picture and sound are
related, but for which there is no need to maintain perfect sync;
`mute' (silent) film, for which no useful sync sound could be
recorded;
library film for which suitable sound may or may not exist.
Film sound available to the editor includes:
.
.
.
[19.23] Identification marks on film
trims
A, Picture trim: 1, Edge number (and
bar-code on extreme edge above it)
on picture only, printed through from
negative and used by neg-cutters
working with a cutting copy to
identify the original.
B, Magnetic sound trim: 2, Code
number (`rubber number') printed on
edge of both sound and picture of
synchronized rushes. Numbers
repeat every 16 frames, so sound and
picture can easily be resynchronized.
3, Shot number, written on trims
by hand for rapid identification
(not essential if trims are
rubber-numbered). 4, Stereo
sound, cut square to avoid `wiggle'.
.
.
sync sound;
wildtrack commentary or dialogue that will generally be played
only with pictures that do not show the speaker's mouth;
wildtrack effects that will be synchronized to action on mute film
(e.g. a gunshot) or sync film for which the effects recorded at the
time were inadequate (e.g. too distant);
Location atmosphere tracks;
Library sound effects, atmosphere or music.
The film editor (or an assistant) synchronizes the rushes or, if this
has already been done, checks sync. To synchronize film, run the
picture and find the frame where the clapperboard has just closed.
With a wax pencil, mark this frame with a boxed `X'. On the roll of
transferred sound, mark the corresponding frame with three bars,
`III', on the backing with a thick felt marker pen (wax can stick to
the magnetic coating of the next wind). Write the shot and take
number ahead of these marks on both picture and sound. For continuous replay the spacing is then adjusted so that this picture and
sound follow in the same synchronization as that of the preceding
shot on the rolls of rushes. Intervening mute film shots and wildtrack
324
Film and video sound
sound may be cut out and wound on separate rolls: all good takes of
sync material can then be seen in the order in which they were shot.
In order to identify trims of sound and picture at all later stages of
editing, the paired rolls are sent to be run through a machine that
inks corresponding serial numbers on the edges of both film and
sound backings, so that they fall between sprocket holes every 16
frames.
At this stage the director and editor may view the film together, so
that the content, intention and cutting style can be discussed. Meanwhile, the assistant logs the material and breaks down rolls that have
been viewed into single shots or related groups of shots and rolls
them up with picture and sound, if any, together. These are kept in
large film cans: each can contains a substantial number of small
rolls, each marked with its shot number, and the cans are labelled on
the edge with the shot numbers they contain. The assistant looks
after this rough filing system from which material can be taken,
restored and retrieved as necessary.
Up to here much of the work can be done on a synchronizer. Most of
the subsequent rough and fine editing is carried out on the flatbed
machine, with the synchronizer used mainly for identifying additional shots and their sound. From this stage on, the general principles and techniques parallel those described for digital editing.
[19.24] Magnetic film trims
C, Mono sound (cut at an angle)
identified by its `rubber number'.
D, Stereo sound trim with no rubber
number printed on this short
segment: the spacing attached to it
shows where the nearest number
(written in by hand) would be.
Chapter 20
Planning and routine
Before any sound or television studio session, some planning is
likely to be necessary ± if only to ensure that all technical facilities
that might be needed are available and ready for use. If a script is
provided in advance, note and roughly mark up the parts that directly
affect you. For television there may be a planning meeting to
co-ordinate what is expected from all the technical and design
departments. Later, in the studio, there are routines for line-up,
cueing, timing, and completing documentation, all of which are
essential to its efficient operation.
These preparations and procedures apply not only to radio or
television, but often also (with minor changes in terminology) to
record company recording sessions and to the film and video sound
dubs (re-recording sessions) described in Chapter 19.
Planning: radio or recordings
For radio or other sound recordings, when there is a range of studios
available the request for a booking is generally accompanied by a
brief description of its use, e.g. the number of speakers or musicians
and type of balance required, such as `discussion, 6 people' or
`novelty sextet with vocalist'.
Except for the simplest programmes, the production office prepares
a running order or script that (except for the most topical material) is
sent to the programme operations staff. Many productions follow an
established routine, so little discussion is necessary before meeting
in the studio; however, if there is any doubt, check in advance. For
example, before recording an unfamiliar musical group, the balancer
should find out what instruments are expected and what sort of
balance the producer wants to hear; if, for example, it were required
326
Planning and routine
to sound like an existing record the balancer would listen to it. Is
the music to be featured, or for background use? Should it be
brilliant and arresting or soft and undisturbing? Will it be heard with
voices over or with announcements at certain points within it? Will
there be applause?
The recording arrangements might also be discussed: its probable
duration, whether it will be stop±start or continuous take, the medium
and number of copies; and arrangements for and time of delivery.
Any unorthodox or experimental layout would be discussed with the
musical director in sufficient time to allow for the provision of
chairs, music stands, acoustic screens and other furniture, and studio
instruments such as a piano. The balancer also sets up microphones
(or checks their layout) and arranges their output on the console,
working out and plugging up (patching) a suitable grouping of
channels, and checks that equalizers, compressors and limiters are
ready for use; that reverberation and other devices are switched to
their appropriate points; that talkback is working; and that foldback
or public address levels are set.
Television planning
For a complex television production the team might include an
operator to replay records and recordings and boom operators
(if booms are in use). Second in command and the most versatile
in duties is the sound assistant, who often leads the sound department's
forces on the studio floor, supervising the rig for musical productions,
distributing radio microphones for a game show, operating the main
boom on drama, or sitting alongside the supervisor to handle the
audience microphone pre-mix, public address and foldback for a
comedy show.
Representing the facilities that the team can offer, the sound supervisor attends the planning meeting (several if necessary) called by
production to co-ordinate design and operational plans. For even a
simple programme, this may involve the director, designer, lighting
supervisor and technical manager (in charge of technical services
other than lighting and sound). More complex productions may
also require the presence of the floor manager, chief cameraman,
costume and make-up supervisors, and others who deal with house
and studio services such as audience management, the assessment of
fire risks, allocation of dressing rooms, and any special catering due
to breaks at unusual times or the arrival of distinguished guests. The
sound supervisor will then ensure that all non-standard equipment
and facilities are booked and available, and that microphones and
other equipment are set ready for studio rehearsal on schedule.
Planning meetings also serve a secondary purpose: as a forum where
ideas can be proposed and discussed, then adopted or abandoned. The
director should listen and arbitrate rapidly between conflicting aims:
these must be resolved in compromise and co-operation. Sometimes
Planning and routine
327
the director modifies the original plan, to make things easier for all.
But if, in a complex situation, the director does not do this, it is often
the sound supervisor who is left with the trickiest problems to solve:
many television directors think picture comes first and sound second.
Sound supervisors who make light work of awkward problems
are therefore valued members of any production team. Equally, those
who accept unsatisfactory compromises and then explain after the
programme that poor sound was inevitable will not be thanked.
At the planning meeting the sound supervisor starts to build (or
re-establish) a working relationship with other members of the team
that will carry on throughout the term of the production.
The television director
The sound supervisor meets various kinds of director (and producer).
Those who outline their needs concisely and precisely and then stick
to them are easy to work with. They will budget for adequate staff
levels, or ensure that time is allowed for any overlap of function;
conversely, they are unlikely to waste others' efforts or demand
extra staff and resources that will not be needed. Other directors
may have few of these obvious virtues, but still b