[sdiy] Digital synthesis using statistical functions?

[Ben Lincoln]

For the last year or so, I've been working on some image-processing 
software, and one of the things it does is statistical processing on a 
per-pixel basis through a cube of multispectral image data. At some 
point, I started to wonder "what would happen if I ran audio through 
these functions?"

It's been quite awhile since I was heavily in touch with the synthesis 
world, so I figure maybe someone is just going to tell me that I've 
"reinvented Steinbrenner-Kreutzman synthesis", or words to that effect, 
but I did a quick web search and didn't see any immediate evidence of 
other people using this technique.

I've just barely scratched the surface, but some of the results were 
interesting enough that I posted sample mp3s and an extremely basic 
Windows command-line utility for performing the synthesis here:

http://www.beneaththewaves.net/Projects/Statistical_Audio_Synthesis.html

Is this already a well-known technique? I'd love to see modules for it 
in something like StudioFactory, where the inputs could be controlled in 
realtime.