Procedural Generation

[k brown]

Perhaps only tangentially related, but I'm reminded of the surprisingly complex results that were produced by very simple means in some of Brian Eno's early Ambient Music. Perhaps most famously '2/1' (or was it '1/2') from 'Music for Airports' where he simply created about seven loops of tape containing a very simple, soft sound; each a different carefully selected pitch (one dissonant) and of non-exact fractions of each other lengths. Once running, the manner in which the notes combined never occurred the same way twice; sometimes quite isolated in time, sometimes in dense chords; the longest loop contained the one dissonant note, so every now and then - spice. I actually created a project with SM that replicated this system, so one could experiment with different pitch sets, timbres, loop lengths, etc. I should flesh that out/refine it more. He called this type of idea 'Generative Music' and it's what led eventually to him using the KOAN, etc.

[Spogg]

That’s interesting Kevin. When I was mining the old SynthMaker stuff I didn’t see this, or maybe I did and it didn’t work in FS. Some stuff didn’t.

So it would be very interesting for this to emerge again. Anything to keep you from painting eh?

Cheers

Spogg

[tulamide]

POnce running, the manner in which the notes combined never occurred the same way twice; sometimes quite isolated in time, sometimes in dense chords; the longest loop contained the one dissonant note, so every now and then - spice.

Reduced to some barebones as is typical for today's era, it came back into modern music in the form of polyrythm, which is playing in two or more totally different signatues, but the same time length. Leads to similar effects.

https://youtu.be/U9CgR2Y6XO4

And here is an example of full blown procedural generated music composition (the procedural part inspired by quantum physics, so we won't compete with that):
https://youtu.be/MiJd7LAZMd4

[k brown]

That’s interesting Kevin. When I was mining the old SynthMaker stuff I didn’t see this, or maybe I did and it didn’t work in FS. Some stuff didn’t.

No, it never 'emerged' in public. Did a couple of versions, one with simple 2-op FM 'generators'. I can throw 'em up here if anyone wants a play while I squeeze in a brush-up on the thing. These were done long ago, and full of bone-headed construction I'm sure.

Be sure to read the PDF on how to set the thing running - before I figured out how to do one with free-running LFOs, the thing just farts out all the notes at once in a frightening blast, before the loops start offsetting from each other and getting down to Ambient business.

An interesting side note, once I got one of these working, no matter what other pitches I tried with it, nothing was as pleasing as the ones Eno used (I think from a mixolydian scale with added dissonant note). The man knows his stuff - not just tech-wise, but musically.

Here's a track I did with, I think the FM version:

2_2_kb.zip

(5.31 MiB) Downloaded 963 times

Only effect used, my long-time favorite - Charsiesis

For those who are curious - the structure of the 1/2 track on 'Music for Airports' is:
18 sec loop - a b'
20 sec loop - f'
17 sec loop - e b'
31 sec loop - d b'
21 sec loop - c'
18 sec loop - a b
25 sec loop - f
Approximately a Lydian scale.

It sounds like the track 2/2 is the same, but with loops of about half those lengths.

https://medium.com/@metalex9/introducti ... e00e4dba11

[Spogg]

I had a play with the “Two-Two fm freeLFOs” schematic and I was really impressed.

As far as I know nobody else has made such a thing with FS and I really do think it’s worthy of more attention. For example, you now have the luxury of having Martin’s wonderful low CPU oscillators, so you could possibly have more voices.

It would also be nice to have an on-board preset manager visible and we could then share presets. And we could go really crazy if we had a choice of sound generator systems for each voice.

Can you imagine an ambient background with the occasional wave folder or noise wash sweeping in and out?
I’m getting excited now…

Cheers

Spogg

[k brown]

It's funny, it was just a week or so ago that I thought of this thing and how much more I could do with it now than I knew how to back then. One thing I wanted try was Martin's Spaced Partials osc that he posted some time ago; it produces a surprising range of timbres from just a single piece of code. Noise - never thought of that, I was so fixated on trying to replicate Eno's piece at the time. Speaking of noise, did you ever try a synth by Majken called Chimera; it's only sound sources are noise gens pitched by highly resonant filters with 100% keyboard tracking - produces some truly unique and beautiful sounds; be cool to incorporate something like that in this thing.

One of the more interesting results I got from this was doubling the length of the loops, then feeding it into a feedback delay set up to approximate Frippertronics.

Shouldn't take too much to flesh this out and polish it up.

Thanks for having a look at it.

[k brown]

Just setting about the re-do of this I realized that, at least for replicating it's namesake, the MFA track 2/2, this is way more complicated than necessary. 2/2 consists of an identical synth tone (ARP 2600) that was recorded at 7 different pitches and on different lengths of tape loop. So for the plugin, only a single filter and single set of waveform controls for all 7 oscillators is needed; or in the case of the FM version a single set of modulator/carrier controls setting the same tone for all 7 sets of oscillators. Of course, the more complex version will be good to have too, when you can spare the CPU and want to do crazier variations.

[Spogg]

Yes, I guess it’s about whether you want just different notes and/or different sounds, the latter being far more complex but with much greater possibilities.

Whatever you do, I’ll be watching you.

Cheers

Spogg

[tulamide]

Whoever comes late to the party, won't understand the original intent of this thread. So please don't be offended, when I summarize quickly:

Please leave a note, if you are interested in learning about procedural sound generation and getting some code that will help you on the way. Procedural generation is very different to generative music, it is more like a machine-supported, classical sound design. Your users might be interested in simplified sound synthesis.

[Spogg]

I’m still VERY interested in anything you might come up with tulamide.

Maybe Kevin could start a new thread for his generative music…

Cheers

Spogg