|
wecouldcrash, the analysis return pitch information on every "segment" in a track. A segment is defined by an onset and a decay. Pitch information is defined by a vector of 12 pitch ratios from 0 to 1. It's not by any means a voice pitch tracker, and you wouldn't know anything about the octave range. If you used monophonic voice tracks, you could try to segregate with timbral information, although I believe it is not really optimal for this use case either. The track analysis API is really for song structure description in terms of rhythm, timbre, and harmony. For more information, have a look at the documentation.
|