The idea was to make a program that would read [.wav] sound files and give you some manipulation tools to extrapolate and export a usable Fcurve for Softimage|XSI. The first step was to make a draft of the program that would read the waveform and export the Fcurve.
A list of things to do:
1) Extrapolate negative samples
2) Implement FFT or some other algorithm to get a smooth curve
So far there's a very crude way to manipulate the waveform, you can select the samples[or keyframes] / second and you have an "amplitude" filter.
My friend Stavros Karposilos who programmed this for me is very busy lately ...so it came to a halt unfortunately.
You can download the draft here [it crashes or exports blank files sometimes]
Here's a screenshot:
And a rough sketch to give you an idea of how the fcurve should be extrapolated from the waveform: