>I wish we could have some wavelet analysis/resynthesis tool, similar to
>that already implemented which uses Short-time fourier transforms
>(pvoc).
>Let me add I think we have now enough computer power to be able to call
>the analysis routines from within the instrument code itself, so for
>Josep M Comajuncosas

>I'll be getting back to work soon on convovle and pvoc units that do the
>analysis from within the instrument. The pvoc works already in my version,
>the convolve might be done in a month or so.
>RK

I dont think it would be harder to encode FFT than it is to encode a wavelet 
instrument. Both cases use a convolution and perform the computation in 
place. Using the FFT/Wavelet code at Numerical Recipes (in case anyone 
missed it, its FREE online)  its easy to port to csound. The expected 
computational effort would be similar to wavelet's as well. I just choosed 
wavelets cause their more amusing, and seem to be well suited for describing 
complex signals like sound, due to its multi-resolution analysis.
I can send a conventional wavelet instr (in a couple of days...), that lets 
you apply the FWT (fast wavelet transform) to an audio signal, and the IFWT 
(inverse-FWT) to the wavlet coefs, to reconstruct it back. Much of it, as 
well as the programming philosophy, can be used in the design of a FFT 
instr. Of course, a couple of dedicated ugens is much better (altho maybe 
not so educative)
Btw, I totally agree our computers can very well deal with FFT computation 
_inside_ csound, and am looking forward for the new ugens