Generative processes & creative intentionality

I have been taking a different approach to ‘generative’ than a lot of people.

I use sequencers like chained Squonks, and use randomize on sequencers as a first step. I then tune the sequence - adding notes, removing notes, re-pitching notes. Randomly generated sequences are the raw material. I listen and say “I do like this, I don’t like that.” I approach the composition from a random starting point but spend a lot of time making it into the best possible version of the initial randomness.

The random sequence is the ‘seed’ which triggers my mental/creative process, but I might end up with something completely different. The random sequence is like a random number generator seed value. Instead of seeding a random process, it’s seeding my creative process.

It saves me from the ‘blank page’ problem of creating art. I’m starting from somewhere. There’s the old cliche “if you don’t know where you’re going, any direction is fine.” I start from something but I really don’t know where I’m going, just that I’ll know when I get there.

I also start with empty sequences and manually add notes, which is a different process. It might be ‘better’ with respect to being a pure product of my creative imagination. Whether the results are better than starting from a random sequence, I don’t know. The same creative process happens: I’m trying to fix something that’s broken in an interesting way. It’s just that’s what’s broken about an empty sequence is that its empty.

What makes music music and not just noise is intentionality. Your unique human intention when you pick up a tool is what a listener can hold on to. It is what gives your work meaning. Your participation.

You can tune a purely generative process so that it reflects your intentions. But in the end that is more frustrating to me, because purely generative systems don’t often have the right ‘handles’ to grasp in order to impose your compositional intention.

I also use a LOT of random modulation; sometimes having 20 or 30 sample & hold modules generating random modulation. THAT is generative, but when combined with more INTENTIONAL processes as describe above, it can be a way to generate surprise in music without it feeling arbitrary. The intentional, tuned aspects of the patch are something the random modulations ornament.

I spend a lot of time tuning the magnitude and center. My go-to sample & hold is the Bark S&H/T&H, because it’s polyphonic. It’s controls map perfectly onto the magnitude/center paradigm, with the scale and offset knobs.

As an example, use a Bark S&H to modulate filter cutoff. If I use bipolar modulation, you control the center frequency with the filter’s cutoff knob and the range of modulation with the scale knob. For mod inputs without attenuverters, you can do the same thing with the Scale and Offset knobs on the S&H.

A low pass filter with too low a cutoff frequency is effectively silent, so your modulation needs to be scaled/offset to where the minimum cutoff is at a frequency where you still hear the signal. Similarly, you may not want the modulation to raise the cutoff to where you get the unfiltered signal. You want the upper cutoff frequency to be musically pleasing.

For the Bark S&H/T&H I save presets on it where I have set every knob for every voice has particular values. My most commonly used are ±10, ±5, 10, and 5. Meaning that every voice in the polyphonic output is in the same range.

The Scale/Offset knobs of the random modulation generators are ‘handles’ whereby you can control how much randomness you want in your output. You create a network of generative sources, but you impose your compositional intentionality on the network by fine tuning the ranges of all those modulations.

When you find the sweet spot you have something pleasing, it will do surprising things that reinforce the artistic intentionality of the deterministic system it modulates. I find myself listening to the pieces I make, and after a few times, the places where something unique and surprising happen don’t feel arbitrary, even though it doesn’t have the pure intentionality of scratching out notes with a pencil on a blank sheet of music paper.

In the end, everything is an experiment and I throw out a lot of failed experiments. But having spent AT LEAST my 10,000 hours working with Rack over the past several years, I fail less and produce more.

At least I like what I make, and I only achieved any level of productivity when I stopped caring about what other people might think. What I have found, based on feedback from listeners, is that when I try really hard to make something that I want to hear, there will be some people who also want to hear it.

Some people in commercial music are really good at making things that many people want to hear, at least the first 50 times they hear it. I can admire that level of intentionality and creativity but often, they don’t create anything durable. A whole lot of pop music blows up, sells a lot, and then gets played out, and people forget it. I’d like to aim a little higher even if it means it limits the audience.

It means my music doesn’t please everyone. But if you try to please everyone, you end up not pleasing anyone very deeply. Better to make the music you really want to hear, and see who agrees with you.


Maybe not so different. I often use randomnes the same way you do and I suspect lots of other people here do as well.

I agree and think the same way. I don’t actually think the word “agreement” plays much of a role in music, it’s more … inspired, moved, affinity, interested, curious, or even … baffled.

1 Like

Thank you for this. I use different techniques, and I get different results, because I am not you. But in general, as Lars said, my approach has lots of similarities with yours.

For that sort of thing (if I understand what you’re getting at) I have recently discovered the DUFFY module.

You can dial in the number (integer) of semitones for the shift, and choose to transpose either up or down.

I like to load up a sequencer (particularly IONS) with some random values, and have them repeat several times before changing just one of them at a time. (CV-MAP is my friend.) In addition I might throw in DUFFY with a Bernoulli gate on the Down and/or Up inputs, so that the pattern stays somewhat familiar but modulated. The beauty of DUFFY is that it wraps around after 12 semitones, so it’s never going to turn into melodies fit only for bats and whales.

Totally doing this lately. I use tools that have randomness sewn into the fabric of the tool, but I just use that part to jiggle up the notes. Ultimately I am selecting a proper version from the possible selected parts. I like that it lets me also visit these little sections again and rejiggle the notes until I am further satisfied. Some favorite sequencers lately have been Slips, Proteus(with the great new expander that saves 5 patterns), and HorseADoodle. HorseADoodle is really fun in that it can do melodies, chords, or even counterpoint. Great subject, turning these random bits into fixed masterpieces still keeps some of the patina of this egoless patching with just a bit of musical intuition into the mix to make it a song.

1 Like