[sdiy] granular synth on AVR or Ucontrollers??

[Tom Wiltshire]

I've tried some of these ideas on dsPIC. I wrote alternative code for a prototype unit sat on my desk and turned it into a granular oscillator.

Some of the criticisms are fair. There are a lot of potential parameters, and it is hard to make 'musical' sounds. I think granular synthesis is probably best done when you've got decent length samples of real-world sound to work with. Single cycle waveforms don't give a lot of scope.

However, you can get some good sounds using windowed bursts of sine or other waveshapes. I used basic triangle windows, since that's simple to generate. I had independent control over the length of the window, the frequency of the underlying sinewave, the frequency of repetition of the bursts, and the amount of randomness applied to each of these variables.

The resulting sounds vary from reasonably controlled bandpass effects, to rainstick-type noises and mechanical groans and sizzles, as Olivier describes.

I'd like to have a faster processor and a big sample to play with this properly. Then you'd be able to do more sophisticated windowing, and be able to move the read point throughout the sample to get timbral effects. Like Olivier says, cSound or PD or such like is probably the way to go for these experiments, at least initially, although in my case I'd rather code it and have a result today than sit down and find my through more documentation to be able to make PD work tomorrow.

Tom


On 5 Apr 2012, at 09:16, Olivier Gillet wrote:

> Granular synthesis from samples on a modest MCU is possible - the WTPA does it.
> 
> Auduino is labelled "granular synthesis" but it really is a variant of
> the CZ-101 style filter emulation
> (http://en.wikipedia.org/wiki/File:Casio.CZ101.resonance.png) or FOF.
> 
> Your idea would work with certain conditions... Especially for 2 you
> need a smooth envelope to "cover up" to avoid the high frequency
> content when a sine wave is brutally turned on / off at the top or
> bottom of its cycle. The biggest challenge is to come up with a set of
> parameters that has an interesting range. If you start offering
> controls for burst frequency, burst randomness, carrier tone
> frequency, carrier tone frequency dispersion, carrier tone
> waveforms... you end up with a large parameter space but most
> combinations are totally worthless.
> 
> You can get interesting sounds by triggering "windowed" bursts of sine
> or triangle waves, with some probability law dictating the dispersion
> of the sine wave from a central frequency ; and the regularity/density
> of the bursts. Can make interesting "noises" - rain, wind blowing,
> dark machine noises - but nothing I've heard of that that couldn't be
> achieved with traditional modular synthesis techniques (noise through
> high-Q filters, noise sent to VCA, noise -> S&H ->  low-pass filter).
> 
> In any case, before ever thinking of MCU, etc, try your idea with a pd
> patch or a csound orchestra to make sure it is musically relevant and
> has the right parameter space.
> 
> Olivier
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy