Generative Patch

Hey everyone! I have been trying to put together something that sounds good as it morphs and sorta decides the notes by chance, to see if I can make something I like. I have been really impressed by the patches I have seen demo’d in the youtube, Vimeo, and Audulus library, and I have spent hours staring, tweaking, researching, and no matter what I do, it seems like I am missing a piece or missing some creativity with my end result, as I have HATED the outcome of almost every patch I have spent hours building and adjusting.

It makes me frustrated enough that I typically will wipe the slate clean, look at how much time went into what is now a blank slate, and I will give up, feeling downtrodden and uninspired. Can anyone provide me with tips about what basic ingredients sound best or give me a suggestion about a good process that will be fun, at least some things that work for you, which I might be able to pick and choose some pieces of that process and perhaps feel a bit better about my outcome? Does anyone else ever feel this way when trying to create something from nothing?

1 Like

I try to deliberately connect things that I wouldn’t normally connect when I’m trying to make something original, or I might split the signal flow into a number of modules then haphazardly sync them up again (maybe using a clock divider or two to retrigger modules at different points in a cycle).

For me the trick is to not think it through too much. Sure, I get a big mess that doesn’t work sometimes, but I’ll often just delete out parts of what I have done and try to insert something new rather than scrap the whole thing.

1 Like

Here’s something I put together from just randomly hooking things together as I described above, and then refining as I went along. It took some wrong turns, and I ripped pieces out and replaced them, but I listened to what was going on more than anything else, and when I started to get the percussive sounds to my liking and the synth sounds were getting interesting I stopped, recorded some of it then processed it as I described in the thread:

1 Like

An oscillator provides the basic material for shaping into sound when dealing with subtractive synthesis. Of course, oscillators – by definition – are periodic, repeating events.

Ok. Fine.

If you stick to the subtractive technique of synthesis, creating generative patches is simple. One approach is to modulate various parameters by using an lfo or noise as a source, then hooking that up to a sample and hold. Even with this very basic technique, the act of describing it is already becoming quite complex. There soon become many layers of modulation.

One thing is simple though – that is, how does a sample and hold work? I’ll let you look into that, but if we take an lfo and have that low frequency oscillator control the frequency of another oscillator, we get variation in pitch. The sample and hold gives us a starting point for sampling pitch so that we get an abrupt change and a steady frequency, which almost sounds like notes. We can then round those frequencies to the corresponding scales by using quantizers.

So, here we are taking semi-random and random sources, but then confining those behaviours by limiting or adjusting the parameters. A very common approach with the sample and hold is to connect a clock signal which would make the sample and hold module output the frequency of the lfo, on the down beat (the clock tick).

Generative Patch.audulus (192.7 KB)

After that base you then look for nice little tools people have made that give you more balance between random and repeating events. The Turing Machine is a step past the sample and hold technique above. Maybe mess around with QRST. A few more steps and you are in trade secret territory – which is just to say that style comes from within.


Here’s something I just had fun cooking up.

Generative Two.audulus (1.1 MB)


Moving to a less technical response, I feel this way all the time in various creative endeavors. Not just Audulus. Sometimes it’s difficult for me to throw paint on a canvas and find something I like. I work much better having an end goal in mind.

I also think that the main reason for this type of frustration is self doubt. The line between dumb and cool is thin, and probably defined by our own confidence. To escape this, I’ll set an end goal in mind based on what I think someone else would make, or like. From an Audulus perspective, I’ll pretend I’m a module designer and ask myself what the end user would want to hear, and how would they want to control it. What empowers them?

Putting parameters around your creativity allows your mind to focus, and not get overstimulated.


Yes I agree. There is a great mystery (at least for the me I take care of) about creativity. The great accomplished dead disagree. Only you know yourself. Then again, science is handy too.

There seem to be two common responses to hangups. 1. Who’s fault is it. 2. What’s the problem.

(Keep in mind that now we can only poke at the debris, chances are the initial interest is a complete write off at this point).

The insurance adjuster cares who is to blame. The mechanic does not. If you want to solve the problem, fix the engine, you need to find a comfortable place where you can work on it with your own mind. The bits you win with your own mind are worth more than getting someone else to understand it for you.

That’s it. Then you connect more with what you make/fix – you take more joy in your understanding it. However, you will encounter yourself. Which is probably the point.

  • There is something else. Having reread the original post… I am interested in Audulus because of what I needed it for. The aesthetic angle is also a must, but I see it primarily as a tool. In a sense, like meditating it is a bit of a practice. Just exploring, not trying to come up with something – knowing, of course that is what you really want, but fooling the mind along until that wee hour. Is it good enough?

Moving along further into creativity, frustration and putting the product ahead of other people’s ideas about how you should be…


That was really inspiring and encouraging; thanks for sharing that! :slight_smile:

One thing I’ve started doing is use a random source that has some center-seeking tendency, aka a bell curve or normal distribution, to give random values or tones a musical flavor. Stschoen has built a Normally Distributed Random Values patch which lets you set the average and the variability for the random values it produces. I built a knob- and bar meter-laden UI for my purposes.

The other thing I’ve done a lot is look to build reusable blocks wherever I can.


All this a scale rotation I’m playing with lately was me trying to replicate a certain sound I love—the Stevie Wonder lilting synth sound from his classic period.

In his song “Too High” from Innervisions, at several points in the bass riff he appears to modulate by half-steps. The riff appears to be played in a pentatonic minor scale.

I already had my bit-rotate module for rotating scales to follow a progression and have all the notes played in that step in the progression remain in-scale, but in this case I tried it to step out of the scale altogether. That the harmonic lift fit so well was unexpected, and it seems to replicate that unique sound in the Stevie songs.

I’m learning my music theory mostly by pushing out what I can do with guitar and by experimenting with Audulus. (I’m no shredder, however! Bossa is more my speed, and speaking of speed, not super fast—I’m not a very fast player!) I don’t have any formal musical training, so I don’t have all the lingo down. It almost seems like a property of pentatonic scales that you can rotate them as I did and they still mesh melodically. Is a there theory out there about modulating whole scales?