Re: [Audacity-devel] Source Separation

[Federico M. <fm...@fc...>]

Dear all,

> We have been started working on one of the Audacity proposals: Source 
> Separation.
>
> http://wiki.audacityteam.org/wiki/Source_Separation_Notes
>
> We'd like to implement it as a VST plugin. The question is how it 
> should work under Audacity? Is there any solution how to deal with 
> multiple outputs VST plugins in this software? Because the obvious 
> approach is to provide different sound sources signals on plugin 
> outputs. Could you give us any other tips/clues?

This is quite interesting but very difficult, there is no general 
solution and in fact it is currently a very active research topic in 
signal processing. There are some algorithms that solve the problem of 
speaker separation in a conversation based on harmonics and formants. 
The same type of algorithm is used in some software that aim at making 
the conversion from WAV to MIDI, which turns my attention to another 
improvement that deserves to be treated some time in the future, i.e., a 
better MIDI support. Now there is an experimental feature that reads a 
MIDI file.

With Luciano we are currently working on what could be a sort of 
interactive source separation feature based on graphical selection from 
spectrogram. It is already working, but it requires some fine tuning 
before we can release it.

Regards,

Federico