top of page

Chapter 2:
What to Modulate And to Trigger

This chapter is more a systematic list of documented examples than arecipe of how to set up a complete patch.

The aspects of setting up whole (finished) patches are dealt with in chapters 3 and 4.

But making music (any kind and with any kind of “toys”) needs a good knowledge of the instrument.

And getting inspiration, getting ideas, realizing ideas etc. needs the experience of what the instrument (and even only a certain part of it) is capable of.
I´m not going to write about modulators as modulating targets. This aspect is dealt with in chapter 1, and I think I thrashed this matter out quite comprehensively.

One last remark before we start: it is inevitable that there will be thematic overlaps with the subjects of chapters 3 and 4 here and there in this chapter 2. Generative music is a complex matter, and quite often it is necessary to look at the same subject from different angles.

Chapter 2.1:

Of course we can modulate pitch, and simply patching and LFO or anothermodulation source in the 1V/Oct jack of a VCO can hardly be a matter of  this book. But what about patching the modulation source through a quantizer to the VCO and not through a quantizer at the same time, and let a Bernoulli gate decide if we hear the quantized version or the other one, which is not quantized to a certain scale.










The result will be a melody in a certain scale with some “foreign” notes here and there, which don´t belong to this scale (nor to any other scale except for microtonal ones sometimes). The preset “pitchmod_1.vcv” shows the patch according to the graphic above.

If I want the notes generated by the LFO-VCA-Quantizer chain to be shaped by an envelope (e.g. in case I wanted plucked notes), but the the other pitches, which are not quantized, shall go unshaped, I simply add
two VCAs, one of which being modulated by an envelope, the other by the Bernoulli gate (the output for the non-quantized pitches). The gate for the envelope I take from the quantizer. I plug the VCAs into two different channels of a mixer, and let the channel of the quantized notes switch on and off the Bernoulli gate (the other output).
The following graphic shows this patch, and the preset “pitchmod_2.vcv” lets you mess around with it.
The video behind the following link explains and demonstrates both – pitchmod_1 and pitchmod_2.







If we want e.g. the non-quantized pitch glitches to be shorter than the quantized melody, we can introduce a third functional group to the patch:  a second clock LFO for the Bernoulli gate and a CV mixer. The second clock LGO runs at a higher frequency, its square wave cycle is shorter therefore.

The “old” slower running LFO shall be patched to channel 1 of the added mixer, and the faster “new” clock LFO shall be patched to mixer channel 2.

The output of the Bernoulli gate, which opens the doors for the pitch glitch is patched to the CV input of channel 2 (the faster LFO), and the output of the Bernoulli gate, which opens the doors for the quantized melody is patched to the CV input of channel 1 (the slower LFO).

Always when the Bernoulli gate tosses a coin and opens its “quantized-melody output” it stays in that state for quite a time, because the next clock impulse doesn´t come before the end of the (long) cycle of our “old” clock LFO.

But when the coin toss falls to the other output (the glitchy one), it tosses the next coin quite fast, because the next clock signal comes fast (from the “new” and faster LFO).
The following graphic and the preset “pitchmod_3.vcv” and the video behind the following link show this.












Of course we can use patches like this to switch between different (regular) melody lines and a lot more. But let´s look at another way to deal with pitch and melody now.
I use the triangle output and the square wave output of one and the same LFO to feed 2 quantizers with CV. The square wave cv is patched through an inverter module before I patch it into the second quantizer. The output of both quantizers are patched into a mixer, and the mixer outputs the combined signal into a VCO.
Always when the triangle wave sends an increasing CV the square wave is at high level, and the inverter sends a zero-level CV cancelling out the triangle wave. We get silence.

But always when the triangle wave sends decreasing CV, the square wave is at low level, the inverter sends high level CV, and we hear the increasing part of the sequence.
The following graphic shows the patch, the preset “opposite.vcv” represents it, and the video behind the next link demonstrates it together with some variations.












Let´s patch some more interesting variations based on this patch now. Patching a VCA (attenuator) between the right quantizer and the mixer input, and modulating this VCA with a second LFO at a higher frequency than the first LFO leads to short wa-wa effects during the rising phase of the sequence (see preset “opposite_2.vcv”).











There are legions of variations imaginable, so only one more: introducing a second VCO, which gets its 1V/Oct CV directly from the first (left) quantizer, and sending the sound of one of the VCOs through a reverb and splitting the stereo channels between both VCOs leads to a quite interesting counter action between the two pitch developments (preset “opposite_3.vcv”).


Chapter 2.2:

Chapter 2.2.1:

Yes, of course filter. Simple. Common. Boring. No need to say a single word about it any more.
Let me start quite simple indeed.

I take a VCO, feed its audio output into one filter, but modulate all three parameters (cut-off frequency, resonance and drive) with 3 different LFOs running at three different frequencies. I use the triangle outputs of two LFOs to modulate the drive and the cut-off frequency, but I use the square output of the third LFO to modulate the filter´s resonance. This third LFO runs at the highest frequency of all three LFOs. I sent the result through a reverb module to the main audio out.
What I get is a continuously changing sound, nearly typical filter sweep, enriched by the asynchronous drive modulation plus some water drop like sounds here and there caused by the (square wave caused) sudden jumps of the resonence.

The following graphic and the preset “filter_1.vcv” (presets are available in the e-book – see show the patch, and the video behind the next link demonstrate it a bit more.










Even without the LFO-VCO connection the patch brings some quite nice reults (the “water drops” get more remarkable e.g.). And the preset “filter_2.vcv” may well serve as a “base camp” for developing sound scapes for a catastrophic film (with fire fighters and approaching aliens or so – well).
And like with all examples in this chapter 2: in “real live” you will modulate in far more complex ways using the networks and units from chapter 1.
There are quite a couple of specialised filters out there, and it cannot be the matter of this book to talk about all of them (in chapter five I´m going to introduce a few more though. So let me mention only the wide spectrum for formant like filters mimicking voice like sounds. The preset “filter_3.vcv” (presets are available in the e-book – see is an example.

The video behind the following link demonstrates all what has been said about filters in this chapter.


Chapter 2.2.2:

Let´s be honest: there are quite a lot examples of generative music, which get boring after some time (except for the producer of those pieces). And there are even theories out there in the www (wild world of weird things), that there must´n happen too many really different sonic goings on, and the changes shall be only smooth and between quite similar sonic events. Well, I don´t think so. And a well dosed application of wave shapers and wave folders can make a huge difference here, can set the accents, which “keep the listener awake”.

When I said “well dosed” I mean the principle “less is more”.

The following example consists of two modulation paths, one of which fades a wave shaper in and out. The wave shaper itself is modulated by a sample and hold module. The process of fading in and out of the whole modulation path happens quite slowly here. The other modulation patch creates a random melody, and is driven by a second sample and hold module.











And here is the whole block diagram:











The preset “shaper.vcv.” and the video behind the following link demonstrate the patch.

Chapter 2.2.3:
Partials (additive)

We can – of course we can – also modulate the onboard functions of a VCO. Some of them – often called “additive” VCOs – allow us to change the amplitude, the tuning and the phase of individual partials of the generated sound.

I´m not going to explain the matter of additive synthesis, partials, harmonics, overtones, partial phase etc. in details here. This is a matter for filling a whole book.

If you need help to understand what a partial in a spectrum does (or even: what it is!), please contact me. Otherwise I suppose you know at least the basic of additive synthesis.
And in the following example I´m going to modulate only the amplitude of certain partials in the spectrum of the VCO´s sound, because at least that is, what all of the so called additive VCOs allow us to do. In general we can say, that modulating a lot of partials at the same time tends to result in quite chaotic developments of the resulting sound, structures which I´d like to call kind of “destructive” and contraproductive for the whole piece. More often than not it´s a good idea to modulate a certain partial or a certain group of partials together with the pitch of the module (sound), and in the same cycles (but not necessarily in the same direction, both up, both down etc).

The preset “additive.vcv” and the video behind the following link use a quite exotic looking additive VCO, but you can use any other VCO, that allows you to modulate single partials or certain groups of partials (only odd, only even, partial 1-10 and partial 11-20 etc.).


Chapter 2.2.4:

And again: I´m not going to explain frequency modulation and phase modulation in detail here. That matter should be covered in a book of its own. Those of you, who think they need some deeper and detailed information about it may watch my workshop series about it, first part here:

But some advice to make live easier (at least at the beginning) may come in handy.

It´s a lot easier to get useful results when you modulate the pitch of the FM oscillator and the output frequency of the modulating module the same way and at the same time. And be very careful with the strength of the modulation, which is equivalent to the output amplitude of the modulating module.
Again: less is more.

Some FM oscillators limit the modulation strength automatically, but most of them do not.
And a last advice: The feedback of the modulation (which will be a dedicated parameter with your module, not a patched feedback) works nicer and is more useful, when changed in jumps (e.g. coming from a quantizer or a sample and hold unit) or fed in only here and there, than changing it in continuous cycles (coming from a sine/saw or triangle LFO).

Only when modulated with the same phase and frequency as the (quantized) pitch the results can be equally useful.
Well, and here is the diagram of the example, which is also represented by the preset “FM.vcv” (available in the e-book – see and by the video behind the following link.

The number of on-board parameters of those hundreds of different oscillator modules out there is legion. Let me only mention some more of the common ones, like pulse width modulation and hard-sync I´m going to talk about some more of these parameters in later sub-chapters here.


Chapter 2.3:


Another kind of modulating, of changing the whole development of a patch is switching on and off different voices. We can achieve this with a (square wave) LFO-VCA combination or with dedicated switches (e.g. Gray Code modules).



















But a far more interesting way to switch between and merge whole voices (audio paths) is using ADSRs together with the VCAs, and patching the LFOs or the outputs of the Gray Code module(s) into the gain inputs of the ADSRs.








The preset “voices.vcv” (available with the e-book – see ) and the video behind the following link demonstrate a patch like this, with 5 voices, 4 driven by a Gray Code module, and one by a
little LFO network.


Chapter 2.4:

Let me set the structure of this chapter as follows:

Part 1: rhythm is more than percussion plus bass
Part 2: switching versus morphing
Part 3: delaying beats and polyrhythms

I´ll start part 1 with an imperfect patch:

An LFO triggers a sample and hold module, which modulates a non-percussive voice. The same LFO feeds a Bernoulli gate, which triggers a percussion module. What I want is, that both, the voice and the percussion generate their sounds following the same beat (the beat of the LFO), but at random times (the percussion) and at random pitches (the voice).

Here is the graphic:





This patch is used in the preset “imperfect.vcv” (available with the e-book  see

The video behind the next link shows its behaviour (compared to the next one, “perfect” version).

Why is this patch imperfect?

Well, it is not – as long as it´s not a part of a larger whole, of a larger patch. The problem is the LFO. In the moment, when I want to use this patch as a part of the “rhythm section” of my piece, I would have to use this LFO and nothing else as the centre, as the heart of all rhythmic goings on. Otherwise its sounds would rarely fall on a common and shared beat with the rest of the rhythm section and the whole beat.
When it comes to rhythm (even when it comes to polyrhythms – see a bit later in this chapter) even randomly interspersed sonic events should fall against a measure of the wanted/adjusted beat (at least most of the time – see the part about delayed beats).

The improved (more useful) version of the patch looks like that therefore:

Clock modules offer more, better and easier to use functions to synchronize and even to change measures, beats and speed (BPM) than LFOs. The preset “clock_1.vcv” represents the improved patch, and the video behind the following link demonstrate both, the imperfect one and the improved one.

To make the voice and the percussion fall together on the same beat more often (the sample and hold module and the Bernoulli gate work independently so far, and its rather seldom that both sound events fall together) I can insert a CV mixer modulated by an LFO´s square wave:
one mixer channel is fed by the sample and hold unit, the other by the Bernoulli gate, and one of these channels (either the sample and hold or the Bernoulli gate) is switched on and off by the LFO.

When I patch the output of the CV mixer into the gate input of the voice, sometimes the voice is triggered only by the sample and hold module, and sometimes by both, the sample and hold as well as the Bernoulli gate.

The following graphic shows the principle, and the preset “clock_2.vcv” represents the patch. The video behind the following link demonstrate the goings on.













We can change nearly any parameter in a rhythmical way of course. Filter cut-off, filter resonance, distortion, any fx-parameter you can think of. The preset “clock_3.vcv” and the video behind the following link show only one example (of hundred and more possible ones):

rhythmically modulating the delay time.

Part 2: switching versus morphing now.

Well, even in generative music it´s not always about changing everything and all the time.

I´ll talk a lot more about this in chapter 3.

Sometimes it´s about two stable rhythms, which don´t change, at least not for a shorter or longer time. No randomness in the rhythms.
Well, not in the rhythms themselves. What is random is the way, the succession, the “frequency” of switching between these two (or more) stable and unchanging rhythms. I can do this switching to and from using dedicated switch modules, and if I don´t want to hard switch but want to crossfade or blend, well, then I chose a switch with blend functions (or add a crossfader module to the switch).

The preset “blend.vcv” gives an example, and the video behind the following link demonstrates it.

Logic modules are also a great help when it comes to switching and/or combining rhythms. We can use them to take random triggers from different sources (e.g. those which we are already using somewhere else in a large and complex patch) and let them switch between whole rhythms according to their logic (AND, OR, NOR, XOR etc.).











But we can use logic modules to really combine and expand rhythms as well. We can e.g. want an additional voice (bass, brass, whatever) to step in, always when the kick and the snare play together. Or we can want the
additional voice (or another one) to step in only when neither kick nor snare are playing, or when only one of them is playing etc.

The preset “voiceAdd.vcv” and the video behind the following link show this.












Delaying beats and polyrhythms now (Part 3).

I´m going to talk about polyrhythms in chapter 3 a bit more in detail.

So let me say it in a simplified version:

Always when a ¾ measure meets a 4/4 measure, or a 3/3 measure a 2/2 measure we call it a polyrhythm. There are different “families” of these rhythms, but that´s a matter for chapter 3.







Let me describe only one special technique here: delaying beats in polyrhythms. Of course we can delay beats everywhere, we don´t need a polyrhythm to do so, but with polyrhythms it´s more impressive and sounds less than a mistake (as it does e.g. in a 4/4 measure, if we don´t take care).

Things get interesting, when we modulate the delay time while having set a very short (if at all) feedback. The beat, that I sent through this delay seems to tramp around its regular position, which gives the whole rhythm a disquieting touch.
The preset “poly_1.vcv” and the video behind the following link give a short demonstration of it.









Another way to set up polyrhythms is using two sequencers with different numbers of steps per sequence an merge them, fade them in and out or switch between them to and fro. To introduce randomness I can use the obvious candidates (Bernoulli Gate, Gray Code module etc.) to do the switching and blending.

Chapter 2.5:

The number of effect modules is tremendous, and the number of all of their parameters is nearly uncountably high.
And all of them can be targets for modulation and randomisation. Let mechoose only 2 examples from this group of targets therefore: the probably most often used delay and reverb effects.

Delay first.

When we randomly modulate the delay time with short pulses while e.g. a melody is fed into the module, we get short bursts of kind of alien-like sounds, which set (random) accents here and there. And we add even more changes, more development to the piece, when we modulate the feedback with any other wave shape than pulses or squares, and let both modulations run at the same time, but at different frequencies (wave lengths).

A simple pitch development (melody) can became acceptably interesting this way. The preset “delay_1.vcv” (downloadable presets are only in the e-book – see and the video behind the following link demonstrate this.








Next example: Reverb

There are quite simple reverb modules with only a view parameters (e.g. only the decay time and the HP-damp), as well as extremely versatile and multifunctional delay modules with tons of reverb parameters.

I use only the decay time in this next (and last in this sub-chapter) example, because all of you will have this parameter on their reverb modules.
Modulating the decay time leads to an effect like reverb echoes. Not the reverb generating sound itself is echoed, but only its reverb (we get a “corrugated reverb tail). The preset “reverb_1.vcv” and the video behind the following link demonstrate this.

Chapter 2.6:

It´s necessary to talk about randomising envelope parameters a bit more in detail now. In chapter 1 we saw that envelopes can be a modulation source as well, that they can be patched generating cycles of more or less complex structure.
Now let´s talk about envelope parameters as modulation/randomisation target.

Modulating the attack time, the decay time or the release time leads to rather subtle than dramatic changes.
Well, it´s all about time, and nobody and nothing can go back in time (at least not in “normal life – some discoveries in quantum physics seem to show, that under certain circumstances …).

Time that has gone by is “lost” for ever.

So, when I speed up the attack time by the falling flank of a triangle wave (just an example) I always shorten the whole attack time. No matter if the rising flank arrives early enough to slow down the remaining “head space” again.
(don´t forget: a falling CV level means less attack time = faster, a rising CV level means more attack time= slower.)

Once having reached the maximal output level of the module the envelope will inevitably jump to its next phase, the decay time (or stay at or fall to sustain level, if decay is adjusted to zero).
The following graphic shows an example.











We see: modulating the attack time will always shorten the overall attack time compared to the time, which the attack phase would last without modulation. And it´s the same with the other time domain parameters decay and release.
The preset “envelopes_1.vcv” and the video behind the following link will
help you understanding.

When the gain signal and the modulating signal are out of phase, then we get different attack times (or different decay times or different release times) with each new gain, but they all will be shorter than the unmodulated version.
And that is why attack, decay and release modulations are rather subtle (shorter effects) than dramatic most of the times.
The presets “envelopes_2.vcv” and “envelope_3.vcv” and the videos behind the two following links demonstrate some low and medium complex patch examples using envelope modulations.

We have already seen in chapter 1, how to randomise/modulate regular or random cycles producing modules, and that we can modulate/randomise mixer channels shouldn´t be a surprise either. No extra sub-chapter needed,
I think.


Chapter 2.7:

A lot of quantizers offer CV input jacks for modulating (randomizing) the scale and the key of a series of incoming pitch CVs. When we do so, it is (most of the times) a good idea to completely randomize only one of these parameters, and modulate the other one by a (more or less complex) repeating regular cycle of CVs.

And why not feeding this quantizer by another one, which produces only a very few different pitches, e.g. allowing only the notes of a certain chord. This way we make changing the key and the scale (one of which randomly changing) better recognisable even for “ordinary” listeners.

When the unit(s), which deliver(s) the pitch CV for this non-transposing quantizer is a regularly repeating cycle (e.g. an LFO or a network of LFOs), then a relation of integer multiples between the three participating frequencies (see graphic) lets the piece sound nearly kind of “classic with a disturbing touch”.




















The preset “quant_1.vcv” and the video behind the following link show an example and may serve as a”base camp” for experiments of your own.

Chapter 2.8:

Granular sound processing is probably the widest field we touch in thisbook. I´ve written a whole book of more than 270 pages about the matter of sonic grains, and it´s only Volume 1 of at least 3 (2 more to come).


I´ve produced a series of about 20 videos concerning granular sound design (some of which are hours long).

A grain and a stream of grains can have dozens if not hundreds of different parameters, and each manufacturer emphasizes on different one. But there are at least three parameters which we can consider basic (and which can be found with (nearly) all modules): the length of a grain, the density of grain streams and the pitch of a grain.

The direction a grain is played back in is a major parameter as well, but not all modules offer the possibility to modulate playback direction.

Anyway, in the modular world there are not that many granular sound processors (opposed to the world of software instruments, where there are a lot of them). Let me focus on the three main and major and most basic parameters therefore. And for those of you, who still need at least a basic understanding of how grains work the following short and quite simple explanations may come in handy.









A sonic grain is basically a (very short) snippet of audio.

The length of a grain can reach from only a millisecond (1/1000 second) up to some minutes (60 seconds is a common upper border with a lot of modules).
Each grain can be pitched differently (and to different tones than the original audio). The pitch of a granular processed piece of audio is independent from its length/playback speed (the explanation how this is achieved shall be left to another book).

How many (overlapping) grains of the same piece of audio are played back at the same time is described by the “density” parameter. Important to know: grains, which are sounding at the same time don´t need to represent exactly the same piece of audio.














If we have a piece of audio, which is remarkably longer than the average grain length, the position from where a grain is taken is important as well.
There is a huge difference whether a module can work on audio, that is streaming in/through the module (real time granular processing), or if we have to load a certain piece of audio in the module, where it then can undergo our granular processings.

Alright! Enough of theory. Let´s jump into granular tweaking now.

I have chosen the (very good software reproduction) of “Clouds” by Mutable Instruments. And even if I´m not going to explain the whole module (I´ve produced a video doing so – it´s nearly 2 hours long!), I´ll demonstrate the parameters length, density, pitch and position and there modulation using this module.

A last warning: Because granular processing intervenes deep into the heart of a sound, it is easy to get dramatic sonic results – but it is not at all easy to get musically meaningful/useful results.

The presets “grain_1.vcv” and “grain_2.vcv” and the video behind the following link may help you doing some steps of your own.


Chapter 2.9:
Sample (Player)

Modulating and randomising parameters like loop-start, loop-end, playback direction etc. with samplers and sample players is kind of general knowledge I think.

The preset “sampler_1.vcv” and the video behind the following link demonstrate one way to apply sample players to our generative music setup:

The main sounds generating process is interrupted and a pre-recorded sample is played back. The interruption of and the returning to the main sounds is executed by a Gray Code module, and loop start and loop end of the sample player are modulated by an LFO, so that we don´t always hear the same part of the sample, when our nicely ambient sounds are “brutally hushed”.



Chapter 2.10:
Slew Limiter

Let´s make it short: (randomly) modulating the slope of a slew limiter will add some extra sonic colours from time to time.
And when the modulation takes place only here and there, and causes quite steep flanks in the sonic development, it sprinkles the whole sound with remarkable sonic effects.











The preset “slew_1.vcv” and the video behind the following link show an example.


Chapter 2.11:

A comparator does exactly what its name says: the module compares to voltage levels and depending on which is the higher one, and which is the lower one the module reacts.

It may send out a gate signal, or a trigger pulse, or both, it may have more than only 1 output, it may have separated outputs for “higher level as” and “lower level as”. We can understand a comparator as a switch, who´s position depends on the relation of two voltage levels.

So, what can we do with a comparator.

A lot more than we might think at first. Just some examples: Instead using a fixed manually adjusted level 1 we can modulate (randomly) this threshold level to make the module switch in different situations. Or we can use it as a logic module with adjustable levels for logic “high/true” and logic “low/false”. Or we can bit crush a sound by feeding it in as level 2 while we modulate level 1 (“threshold”) by a square wave, and feed the result (from the gate output) to our audio units.









Or we can turn any wave shape into a square wave by feeding it in as level 2 and adjusting level 1 at such a voltage, where we want the outputs switch from low to high.
Or we can use different audio files for level 1 and level 2 to get grain-like effects. Given all the methods of randomizing and/or complex cycling parameters I´ve talked abut in chapter 1 a comparator turns out to be a
quite flexible tool in our patches.

The preset “comp_1.vcv” and the video behind the following link show an example.

Chapter 2.12:
Pitch Shifter

Pitch shifters are meant to, well, adjust the pitch of an audio source, so that it fits into a mix, into its “cooperation” with other audio sources. There are  – like always in the world of modular synthesis – a couple of different pitch shifters around. The video behind the following link shows some of them.

Most pitch shifters allow us to set the amount, the range of shifting the audio source´s pitch measured in semitones, tones and octaves. Others show additionally the exact frequencies etc. These functions make them more suitable for exactly tuning and/or exactly modulating the tuning than e.g. an LFO-FMin combination, when we want to tune/modulate VCOs.
I like to use pitch shifters as tools for moderately changing the pitch of pre-recorded sound files from time to time, as I show in the video behind the following link.

The preset “pShift.vcv” may serve as a starting point for experiments of your own, and the link leads to the mentioned video.

Well, the question “What can I modulate and randomize” could have been answered, even without producing a whole chapter, by simply saying “everything”.

But giving some examples – probably the most important ones – came in handy here, I think.

The next chapter outlines the practical meaning of what I have been talking about in chapters 1 and 2 from the perspective of a composer. It doesn´t emphasize technical aspects, but rather the facets of art and musicality of a piece.

And in chapter 4 I´m going to change my point of view to the opposite, and combine the knowledge of chapters 1 and 2 to introduce technical and organisational “molecules of generative patch organisms”.


… to be continued



Rolf Kasten


bottom of page