Re: [microsound] stochastic waveform synthesis

[Tad Turner <tad@xxxxxxxxxxxxxxxxxxx>]

On 7/2/02 at 7:12 PM, fingerprince <fingerprince@xxxxxxxxxxx> wrote:

> http://www.iua.upf.es/~xserra/articles/msm/stochastic_synthesis.html
> 
> Maybe this helps...

Actually this article describes a synthesis technique that combines
deterministic synthesis with a stochastic component. Really quickly,
just imagine an additive synthesis technique with some noise.

The idea is to more closely approximate the properties of a real
instrument. For example, you could model the sound of a flute with a set
of sine waves at frequencies matched to its fundamental and harmonic
overtones. The result might be a little sterile, so you'd add a
stochastic component to account for the flutist's breath and the natural
physical imperfections of instrument.

At least that's what I got out of a quick reading...

You might want to look at Dodge and Jerse, section 8.1B (pp 266-278) and
the set of GEN21 routines in Csound. I think these are closer to what
Kim was referring to.

Dodge & Jerse's FORTRAN code (pretty simple to follow) demonstrates the
intersection of a set of random variables (the output of a random number
generator) indexed by a variety of equations ranging from Linear to Beta
to create (what I think you'd call) stochastic (or probabilistic)
distributions.

The Csound GEN21 routine will generate these as waveforms that you could
use as samples.

The other important thing to remember is that you can use ANY equation
as an index for a random number generator (this is what the FORTRAN code
suggests), so you can start pillaging chaos theory, etc.

A interesting place to start with some of that is Theory.org. The fellow
that created it is here on the microsound list. Check out the
sonifications!

-Tad
<tad@xxxxxxxxxxxxxxxxxxx>