doepfer
1. Introduction
The A-129 /x series of modules forms a modular
vocoder. ‘Vocoder’ is an abbreviation of ‘voice coder’.
The basic components are an analysis section
(A-129 /1) and a synthesis section (A-129 /2).
Like a ring modulator, the vocoder needs two input signals: a speech element which serves as the raw material for the tonal shaping, and is patched into the analysis section; and a carrier signal, which is patched via the instrument input into the synthesis section.
The speech signal is chopped up and analysed in the
A-129/1 module, and then combined with the carrier signal in the A-129/2 synthesis section. As a result of this procedure, the carrier signal assumes the tonal character of the speech signal, but with its own pitch maintained.
Since the A-129 is a modular vocoder, and the connections between the analysis and synthesis section are external, using patch-leads, you can use this
System A - 100
Modular Vocoder
A-129 /1/2 interface to patch in your choice of modules (eg.
attenuator, slew limiter, CV-to-MIDI / MIDI-to-CV interfaces,, inverter, etc.).
The Five-way VC slew limiter / offset generator /
attenuators (A-129 /3) and Slew controllers (A-
129/4) are particularly designed for this purpose.
There’s also the possibility of connecting the frequency bands of the analysis and synthesis sections arbitrarily, so that, for instance, a low frequency band in the speech signal can control a high frequency band in the carrier signal.
The Voiced / unvoiced detector (A-129 /5) can recognise voiced and unvoiced sections in the speech signal, and switch the carrier signal accordingly.
The A-129 /2 synthesis section can also be used as a stand-alone voltage-controlled filter bank (see chapter 6, User examples).
1
A-129 /1/2
Modular Vocoder
2. A-129 /1, /2 - Overview
System A - 100
Speech
Input
A-129 /1
VOC-A
VOCODER ANALYSIS SECTION
Low Pass Band 5 Band 10
Band 1 Band 6 Band 11 doepfer
A-129 /2
VOC-S
VOCODER SYNTHESIS SECTION
Instrument
Input Low Pass Band 5 Band 10
Band 1 Band 6 Band 11
High Pass
Out
Band 2
Band 3
Band 7
Band 8
Band 12
Band 13
Band 4 Band 9
Control Voltage Outputs
High Pass
Band 2 Band 7 Band 12
Band 3 Band 8 Band 13
Vocoder
Output Band 4 Band 9 High Pass
Control Voltage Inputs
2
doepfer System A - 100
In / Outputs:
A-129 /1
!
Speech In : Input for the speech signal
" CV Outputs : 15 CV outputs with control LEDs
§ High Pass : Signal output from the high pass filter
Filter
Low Pass
Band 1
Band 2
Band 3
Band 4
Band 5
Band 6
Band 7
A-129 /2
$ Instrument In : Input for the instrument signal
% CV Inputs : 15 CV inputs
& Vocoder Out : Audio output from the vocoder
Modular Vocoder
A-129 /1/2
Frequency
100 Hz
120 Hz
160 Hz
230 Hz
330 Hz
500 Hz
750 Hz
1.1 kHz
Filter
Band 8
Band 9
Band 10
Band 11
Band 12
Band 13
High Pass
Frequency
1.3 kHz
1.6 kHz
2.3 kHz
3.3 kHz
5 kHz
7.5 kHz
10 kHz
Tab. 1: Filter frequencies in the analysis and synthesis sections
The following Table 1 shows the cut-off frequency of the low pass filter, the middle frequency of the band pass filters (Band 1 to Band 13) and the cut-off frequency of the high pass filter.
3
A-129 /1/2
Modular Vocoder
3. Basic principles
System A - 100 doepfer
The closer the audio spectra of the speech and carrier signals are, the more speech-like the resulting reconstruction.
The fundamental modules in this vocoder are the
analysis section A-129 /1 and the synthesis section
A-129 /2 (see Fig. 1).
The speech signal is analysed in the A-129 /1, by being passed through a set of steeply sloping band
pass filters, with a low- and high- pass filter mopping up the bottom and top frequencies respectively.
Attached to each of these filters is an envelope follower, which analyses the audio signal level passing through, and sends a proportional voltage out of its dedicated CV output (see below for further details).
The instrument signal is likewise sent through another set of steeply sloping band pass filters, and a low- and high-pass filter in the A-129 /2 synthesis section, and is split into individual frequency bands.
This time, each filter has an associated VCA (voltage
controlled amplifier), which is governed by the voltage present at its CV input.
In this way, each frequency band in the instrument signal has the dynamics of the corresponding band from the speech signal superimposed onto it. The pattern of the speech signal is thus re-constructed from the tonal raw material of the instrument signal.
Speech In
LPF
BPF 1
BPF 13
HPF
EF
EF
EF
EF
Analysis A-129 /1
High
Pass
Out
Instrument In
LPF VCA
BPF 1 VCA
BPF 13 VCA
HPF VCA
Mix
Synthesis A-129 /2
Voc.
Out
Fig. 1: Block diagram of the A-129 analysis and synthesis sections
4
doepfer System A - 100
H In most standard vocoders the voltage signals from the analysis section are fed straight into the synthesis section. With the
A-129 modular vocoder, they are patched externally via 15 leads.
That means it’s possible to modify the control voltages by patching any sensible choice of module (for instance attenuators, slew limiters, LFO, CV-MIDI / MIDI-CV interfaces, inverters, etc.), between the analysis and synthesis sections. Not-so-sensible choices may produce interesting results, too.
It’s also possible to interconnect control voltages to synthesis section inputs in a nonstandard way, so that for instance the output from a low frequency band from the speech signal can control a high frequency element of the carrier signal.
With a modular vocoder, the only constraints on experimentation are the limits of your imagination (and you can also always have a look at chapter 6, User examples).
Modular Vocoder
4. In / Outputs
A-129 /1/2
!
Speech In
Socket ! is the input to the analysis section. This is where the speech signal is patched in.
Don’t forget that the speech signal needs to be at the high level the A-100 uses internally. Plugging a
microphone directly in to the vocoder won’t work.
You need to use an A-119 External Input module, into which you can plug a microphone or other external signal. Then the output of the A-119 can be patched into input socket ! on the analysis section.
"
Low Pass • Band 1 to Band 13 • High Pass
These are the CV outputs " from the analysis section, whose voltages are determined by each filter’s envelope follower. Each CV output has an LED connected to it, showing the strength of the voltage generated.
5
A-129 /1/2
Modular Vocoder
System A - 100
§
High Pass
Socket § on the analysis section is the high pass
filter output. Unlike the other sockets, this is an audio
output, which sends out the part of the speech signal which the high pass filter lets through. This is most usually added to the vocoder output, to make the modified carrier signal more speech-like still, by including these high frequency elements of the sound.
$
Instrument In
Socket $ on the synthesis section is where you patch in the instrument that will provide the carrier signal
(see below).
doepfer
H With an A-129 /5 voiced / un-voiced detector module, you can switch the carrier signal depending on whether a speech signal is present.
%
Low Pass • Band 1 • Band 13 • High Pass
The CV inputs % on the synthesis section are where the control voltages from the analysis section are patched in.
&
Vocoder Out
Output & on the synthesis section is the audio output for the whole vocoder.
P Experiment with different sorts of sound for the carrier signal, for instance
• sawtooth or square waves from a VCO,
• noise (A-118),
• digital noise (A-117).
6
doepfer
5. User examples
Basic principles
To get the best results from the vocoder, it’s essential to take note of the following important points:
• For professional results, the quality of the speech signal is crucial.
If you use a cheap and cheerful microphone, connecting it up to the vocoder via the A-119 won’t guarantee good results.
Any unwanted noise (rumble, airborne background sounds, etc.) will greatly reduce the effectiveness of the vocoding.
According to various musicians including Kraftwerk, the speech signal is easier to use if it isn’t live, but has been taped or sampled, and thus has reliable levels and signal-to-noise - and is repeatable.
System A - 100
Modular Vocoder
A-129 /1/2
• For the best results, speech and carrier signals need to have similar frequency spectra. A quiet female voice, or a child’s, needs a different carrier signal compared with a low-register male voice. If you use a VCO as the carrier signal, you can tune it to find the ideal frequency.
• Basically, the instrument’s carrier signal needs to be as overtone-rich as possible, with a dense audio spectrum. With a VCO the sawtooth output is best suited to the task. An exact square wave has only half as many harmonics, and triangle and sine waves are completely unsuitable (see the notes to the A-110 and/or A-111).
• For professional results, it’s recommended to use a graphic or parametric EQ to equalize the speech signal to produce the most speech-like results at the vocoder’s output. Good results can also be obtained using computer-generated speech (as on the A-100 demo CD).
• For early experiments, radio news stations provide good raw material, because they are nearly always putting out a steady stream of human speech.
• In addition, we plan to bundle an audio cassette of speech with each vocoder.
7
A-129 /1/2
Modular Vocoder
System A - 100
Using just the basic modules
Just with the A-129 /1 and A-129 /2 modules (and an
A-119 external input), all the common vocoder effects can be produced (see Fig. 2).
D First patch all the CV outputs on the analysis section to their respective CV inputs on the synthesis section (band 1 to 1, 2 to 2, and so on)
D Use an A-119 (External Input) to patch an audio signal (see above, chapter 5, Basic Principles) into the speech input socket of the analysis section at normal A-100 operating level.
D Experiment with different audio signals for the carrier frequency (instrument input), for instance:-
• different overtone-rich waveforms from a VCO,
• pink or coloured noise from an A-118,
• digital noise from an A-117,
• ring modulator outputs,
• two VCOs modulated in the audio range by FM or AM.
D Swap the connections between analysis and synthesis sections (see above).
doepfer
A-119
Audio *
* : VCO
Noise
Dig. Noise
Ring Mod.
AM
FM
...
Speech
In
LP
BP 1
BP 2
A-129 /1
High
Out
BP 12
BP 13
HP
LP Instrum.
BP 1
In
BP 2
Voc.
Out
A-129 /2
BP 12
BP 13
HP
A-138
Fig. 2: Basic vocoder schematic
"Frequency displacement"
If instead of patching the outputs from the analysis section to their ‘proper’ respective inputs in the synthesis section, you swap them about instead, interesting frequency displacements occur in the vocoder output.
Fig. 3 shows some simple variations; experiment withall sorts of other possibilities.
8
doepfer
Speech
In
LP
BP 1
BP 2
A-129 /1
BP 11
BP 12
BP 13
HP
Speech
In
Speech
In
LP
BP 1
BP 2
A-129 /1
BP 11
BP 12
BP 13
HP
Instrument
In
LP
BP 1
BP 2
Voc.
Out
A-129 /2
BP 12
BP 13
HP
"Freq. up"
Instrument
In
LP Instrum.
BP 1
In
BP 2
Voc.
Out
A-129 /2
BP 12
BP 13
HP
"Freq. down"
System A - 100
Modular Vocoder
A-129 /1/2
"Chopped up” speech
The patch in Fig. 4 produces chopped-up speech: the vocoder chops speech up rhythmically, in time with the trigger signals. The vocoder output is patched into a
VCA, which is controlled by a rhythmical pulse from an
ADSR (A=0, R=0, D and S to taste). The source of the trigger signal could be an MAQ 16/3, Schaltwerk or trigger from a MIDI sequencer via a MIDI Interface such as the A-190.
Speech
In
Instrument
In
LP
BP 1
BP 2
A-129 /1
BP 12
BP 13
HP
LP
BP 1
BP 2
Voc.
Out
A-129 /2
BP 12
BP 13
HP
VCA 1
ADSR
Fig. 3: "Frequency displacement"
Rhythmic trigger
Fig. 4: Rhythmically chopped-up speech
9
A-129 /1/2
Modular Vocoder
System A - 100
Using with the other modules ( /3, /4, /5)
While extremely usable vocoder sounds can be produced with just the two basic modules, total flexibility and unlimited possibilities are offered by adding on the extra modules (A-129 /3, A-129 /4, A-129/5).
Full user instructions will be found in the modules’ own manuals.
MIDI
MIDI In
CV 1
CV 1
A-191 *
CV 15
A-129 /2 as a MIDI-controlled filterbank
The vocoder’s synthesis section can also, in conjunction with a special A-191 MCV16 module, be used as a MIDI-controlled filterbank (see Fig. 3).
The level of each of the control voltages (Input &) determines the relative level of each of the frequency bands at the A-129/2’s output socket %.
These CVs are patched from a special A-191 MIDI-
CV-Interface (with 16 CV outputs and no MIDI-LFO) and are controlled by various continuous controllers see the A-191 manual for details.
CV 16
* Special version
Audio In
LPF VCA
BPF 1 VCA
BPF 13 VCA
HPF VCA doepfer
A-129 /2
Mix
Audio
Out
Fig. 3: The A-129 /2 as a MIDI-controlled filterbank
10
