I've written some c# code to produce the MFCC from a set of floating
point magnitude values returned from FFTW. Does anyone have any test
data (ie, FFT magnitude values and their resulting 12 MFCC values)
that I could use to verify my code is working correctly? Failing
that, could I pass a set of the 64 values from the FFT for someone to
run through Matlab and give me back the 12 MFCC values? I've been
using this document as the basis for my code...