Fork me on GitHub

The only way to extract pitches from wav sample is to do FFT transform. That data is quite useless if you are looking for absolute pitch/syncing with other pitches, more of an effect with vocoding, you can have one sample pitch bending other sample or audio in.

Petrus Theron19:09:16

OK, can you point me in a direction? I get that there are many conflicting pitches, but my sample e.g. is a just me whistling a few notes. I want to know the relative pitches of those notes, because I am not musically trained. I’ll try to find some FFT extractors.


There are FFT extractors in Supercollider and also Csound, in supercollider its called PV_RecordBuf, don't know if there's a parallel function in Overtone. In Cound it .pvx file and with CsoundQt there's a graphical front end utility to generate these files from a loaded sample. Also, you should be glad to not be musically trained, otherwise you would compose silence for 4 minutes, by which you would get cash from culture and arts fund, and end up finding yourself at the next coffee shop for many hours a day, trying to convince unemployed people around you how great you are. (I say as someone that graduated as composer 🙂 )


Ahh, even easier, the spectrograph analysis tools that come packed with Audacity are very powerful, for example when analysing formants of the human voice. I hope to make a tool that autamates that (one only needs the -6dB spectral peaks, center freq and bandwidth).