High Quality Podcast Recording

High Quality Podcast Recording
Version 1.2, August 25, 2013
Markus Völter
(voelter@acm.org)
Abstract
A good podcast comes with great content as well as great audio quality.
Audio quality is as important for podcast as is layout, style, and formatting
for documents. Just as these parameters influence readability, audio quality
influences “listenability”. If you want to get a decent-sized audience as
well as (potentially) interesting guests you have to make sure your content
is presented adequately. Audio quality is an important ingredient here.
Note that today most well-known podcasts have really high production
value, including professional audio quality. Some of the podcasts describe
the equipment and the recording process, most don't. This document
describes omega tau podcast’s approach to recording decent quality audio .
While it is not as professional as what Leo Laporte does [1], is also much
more affordable. I hope this helps new podcasters get up to speed faster
than I J
Environment
The most important ingredient to good audio quality is the quality of the
raw audio. An important parameter here is the environment in which you
record. Obviously, it should be as quiet as possible. That also means that
you should make sure your computer does not produce excessive
ventilation noise. There are silencer kits, although I don't use any of these.
The next important consideration is the dryness of sound. The sound is dry
if it has no echo or other effects applied to it. For the recording this means
you should try to avoid any echo whatsoever. This is surprisingly hard to
achieve outside professional studios. An office room with lots and lots of
books and carpet and stuff is a good start. The inside of cars is also good!
You can also use specific dampers to achieve this effect. For example, I use
a couple of HOFA audio absorbers [2] to dry up my voice.
More recently, I am also playing with putting an old mattress behind me
when recording.
Apart from just sounding better, not having an echo on the recording also
makes editing easier. It is hard to cut something out when “echo sounds”
remain after cutting.
Recording Equipment
Microphone
The next step for good audio is the microphone and recording equipment.
As for the microphone I use a Heil PR 40 [3] together with a pop filter from
BSW [4]. I also use a suitable microphone arm [5] and a spider web
microphone mount from Heil. This makes sure that the microphone is free
from movement noise from the surrounding.
Voice Strip
The next step in my processing chain is a microphone preprocessor, also
known as a voice strip. We use two of them - one for Nora, one for me:
Aphex 230 [15]:
dbx 286a [6]:
These devices have a couple of features that improve the sound of voice.
The specific settings require some tweaking. It is important to make the
voice sound “full”, but you don’t want to make it a radio DJ-style, big
bottom booming voice either. The following table briefly explains these
features and the settings I use for the dbx 286a.
High-Pass
Removes sounds below 80Hz
On
Compressor
Used to adjust the volume of your
voice: increase quiet stuff,
dampen loud parts.
Drive: 5
Density: 4
De-Esser
Removes the loud hissing sounds
produced for example by ssounds.
Frequency: 5k
Threshold: 6
Enhancer
Basically an equalizer
LF Detail: 5
HF Detail: 3.5
Expander/Gate
Removes very very quiet sounds
and zeroes it.
Threshold: -15
Ratio: 1.5:1
Output
Output Gain
Gain: +8
(make sure it never clips!)
Mixer & Recording
As a mixer, I used the Onyx 1620 [7]. In retrospect, I could have used a
smaller one, but to be able to record two voice channels, Skype, and/or
telephone I recommend at least 4 to 8 channels (I am currently thinking
about switching to a Presonus AudioBox 44VSL).
I use the Onyx FireWire plug-in [8] to feed the audio from the mixer into
the PC on which I record. Note that this device feeds every channel
separately into the PC. It is absolutely crucial that you record every
channel separately to be able to adapt various audio parameters in
postproduction for each channel.
The actual recording is done on a PC running Adobe Audition [13] based
on the Onyx FireWire driver. Recording could also be done with the free
Audacity tool. However, for editing, Audition is significantly better.
Cables
From time to time I hear that good audio cables are important for good
sound. I have to say I cannot confirm this. While I don't use the cheapest
cables, I don't use anything fancy either. You might want to make sure you
don't have any unnecessary adapters or plugs, though: in my previous
mobile setup I had used XLR to 1/4'' stereo jacks and this actually did
produce noticeable noise.
Recording from Skype
If you want to record from Skype, there are various Skype recorder
software packages. However, my experience is that external recording
equipment is better. In my case I simply route the line-out of the notebook
that runs Skype into one of the line-in channels of the Onyx mixer.
Skype can produce very good audio quality if you're lucky and if you stick
to certain rules. All of them are explained perfectly in this video [9]. Apart
from the obvious stuff such as shutting down everything else on your
machine, especially the thing about a dedicated fixed port (which avoids
intermediaries in the packet transport) seems to be very important.
Also, you want to make sure that your guest at the other end uses an
acceptable microphone. 20 EUR USB headset microphones are good
enough, the built in microphones of most notebooks aren't.
Recording from the Telephone
In many cases you will also want to record the good old telephone. To do
so, I use a Telos ONE digital telephone hybrid [10].
It captures "the other end" of the telephone conversation and feeds it into
the mixer. It also feeds my own voice signal from the mixer into the
telephone. The setup produces really good quality if the telephones you
use are connected via real land lines. Voice over IP phones (found more
and more often in corporate environments) will result in significantly
reduced quality.
Mobile Equipment
To record conversations outside of my studio I use a Zoom H-4n [11] as the
recording device. It can record four tracks over all simultaneously: two
external microphones, and two built-in microphones. It creates
uncompressed WAV files. It also has a built in compressor.
More recently I got a Roland R-44 [16]. It has four microphone inputs, so
we can record conversations with more than one guest or one interviewer.
We use AKG C 520 [12] headset microphones will pop filters; they can be
plugged into the H-4n or the R-44 via the XLR connectors directly.
Using headset microphones is important to make sure the speakers don't
move the microphone away from their mouth while talking. Make sure the
microphone does not touch the speaker, and put it outside the mouth/nose
breathing airstream.
Postprocessing
If you record the raw data in the way explained above you will not have to
do a lot of postprocessing (of course you have to do content editing, but
not a lot of sound stuff).
The most important thing I do is adjusting the levels to make sure that the
voice has an even level and is pleasant to listen to, especially in noisy
environments where podcasts are often consumed. Let's look at an
example.
Here is the waveform of
a raw recording. Notice
the different levels
(amplitudes). We want to
make sure they are more
even after postprocessing.
If you zoom in you can
see that the parts that
look quiet really are not
completely quiet (which
will be a problem when
we use the limiter later).
So the first thing I do is
to run the dynamic
processor to remove
everything below -90 dB
(sometimes also below 80 dB if there is more
noise).
Alternatively you can
also use a noise reducer,
if it is good quality (the
Audition noise Noise
Reduction (Process)
works well.
As you can see, this
makes quiet parts really
quiet. This also makes
editing much simpler!
I then use the normalizer
to "pull up" the max
amplitude to -1 dB,
exploiting fully the
available dynamic range.
I then use the limiter to
cut away everything
above a certain dB limit.
Depending on the raw
material this can be
anywhere between -3
and -15 dB.
I then normalize again to
-1 dB. This results in a
relatively homogeneous
amplitude. In the
example at the right you
might want to do the
limit/ normalize
sequence once again, for
example at -4 dB.
To make the editing process more productive, I use a Contour ShuttlePro
[18]. This device (shown below) supports zooming and moving the
waveform and the most important editing operations (ripple cut, delete,
undo) in a much more productive way than the keyboard. While it takes a
while to learn how to “play” this device (I use the Contour ShuttlePro with
the left hand and the mouse with the right hand), you are much faster with
editing.
Mixdown
There's not much to consider during mixdown. If you use the
postprocessing approach shown above for all the constituent parts of the
final mix, the overall volume should be fine (if not, you can always run
Levelator [14] or upload your material to auphonic.com). Here are a couple
of things to consider:
•
Sometimes you want to adjust the overall level between the tracks.
For example, things recorded via the phone aren't perceived as loud
as locally recorded material (phone recordings have less base!). So
you might want to boost the phone track by 2 or 3 dB (or lower the
local track by that amount).
•
If you use music, especially highly compressed rock music, make
sure you reduce it by 3 to 4 dB since it is always perceived to be
louder than the actual numerical dB value suggests.
•
Finally, if you have a dialog of two people, you might want to use 15/+15 stereo panning.
Final Levelling
As a final step, I run the finished MP3 through auphonic [17], a free audio
production web service. It does overall leveling, according loudness
standards used for radio and/or online publishing. It also supports
defining presets that include most of your ID3 Tags, so this simplifies the
overall production quite a bit.
References
[1] http://wiki.twit.tv/wiki/Equipment
[2] http://hofa-akustik.de/pages/produkte/absorber.php
[3] http://www.heilsound.com/pro/products/pr40/index.htm
[4] http://www.bswusa.com/proditem.asp?item=RE27POP
[5] http://www.heilsound.com/pro/products/pl2t/
[6] http://www.dbxpro.com/286A/286A.php
[7] http://www.mackie.com/de/products/onyx1620/
[8] http://www.mackie.com/products/onyxfirewire/
[9] http://www.blogarithms.com/index.php/archives/
2007/12/23/skype-for-interviews/
[10] http://www.telos-systems.com/one/default.htm
[11] http://www.zoom.co.jp/english/products/h4n/
[12] http://www.akg.com/site/products/
powerslave,id,984,pid,984,nodeid,2,_language,EN.html
[13] http://www.adobe.com/products/audition/
[14] http://www.conversationsnetwork.org/levelator/
[15] http://www.aphex.com/aphex-products/973/
[16] http://www.roland.com/products/en/R-44/
[17] https://auphonic.com/
[18] http://retail.contourdesign.com/?/products/23
Appendix 1: Wiring
The following diagrams show the actual wiring of my equipment.
Normal Setup
Telephone Recording