[sdiy] Additive Synthesis - Wendy would say it works...

[Chris Stecker]

> Fourier additive re-synthesis, would be possibly the ultimate synthesis
> method IMHO. Or at the very least, offer new ways of attacking harmonics.
> But despite having the DSP power these days, no-one has really gone there
> to any great effect. They spend their time digitally modeling that which
> has gone before instead of looking into the possibilities that DSP could
> render. And personally, I feel it is THAT which makes digital synthesis a
> second fiddle and not to do with anything that it may sound like.

I think I'd be remiss if I didn't mention that David Wessel and the folks at 
Berkeley's CNMAT (www.cnmat.berkeley.edu) were doing a lot with Fourier 
analysis and re-synthesis from a real-time control perspective when I was a 
grad student there in the 90s. A lot of the work used resynthesis of existing 
sources (modelled strings, winds, or resonances; sampled recordings of 
singers, etc) as a starting point, with the goal of providing perceptually 
relevant controls or "handles" on the data. They used a lot of data reduction 
techniques ala artificial neural networks and things like that. I think 
there's still some of that work going on. Before exposure to CNMAT, I'd never 
thought of additive as very useful, but I heard and saw some pretty cool work 
along those lines. Check it out. 

Chris




-- 

G. Christopher Stecker, Ph.D.
Research Psychologist, VA Research Service
cstecker at ebire.org / 925-372-2000 extension 5073

Human Cognitive Neurophysiology Lab
VA Northern California Health Care System
150 Muir Road, 151-I / Martinez CA 94553