FFT Inverse return not normalized samples about nwaves

[ar1st0crat]

Hi @ToGoOrNotToGo !
Your remark is reasonable. There are at least 2 techniques for FFT normalization:

1. FFT: no change; IFFT: divide by N
2. FFT: divide by sqrt(N); IFFT: divide by sqrt(N)

Hence, I decided not to apply any kind of normalization in corresponding methods leaving it to user (yes, it leads to IFFT(FFT(X)) != X). Normalization can be applied in -Spectrum- methods, though.
I guess, I won't touch currently existing methods Direct and Inverse in Fft class. But I'll add "normalized" methods (just like in Dct classes) to Fft.

[ToGoOrNotToGo]

OK cool. I am waiting for it :-)
I found this:
https://dsp.stackexchange.com/questions/11376/why-are-magnitudes-normalised-during-synthesis-idft-not-analysis-dft

What's your plan with this project? Further developments and/or improvements?

[ar1st0crat]

I've added simple methods under the common name InverseNorm() to all Fft classes. These methods return samples normalized by the Fft size.
As for my plans, I'd like to (1) add 3d audio functions, (2) add more feature related to audio synthesis, (3) add "reconstructors" - objects whose purpose is opposite to that of feature extractors, i.e. reconstructing signals from sequences of audio features.
I'm not quite sure that it'll happen soon, though... ))