Vocal removal/extraction

Archived from groups: rec.audio.pro (More info?)

Sure.
After frequency analysis, with a large number of frequency bins (e.g. 4096),
the phase and amplitude between channels of each frequency bin is compared.
If sufficiently the same, they belong to the centre channel (i.e. mainly
vocal), if different they belong to the rest.
You can throw out either the 'vocal' frequencies or keep them, and then
transform frequency domain back to the time domain.

Matt

"Laurence Payne" <l@laurenceDELETEpayne.freeserve.co.uk> schreef in bericht
news:shu2l0ljl55vfrgnnpqje17gbp42lppfu3@4ax.com...
> For years I've been giving the stock answer - vocal removal - possible
> up to a point on stereo material using phase reversal, vocal
> extraction - not logically possible.
>
> I've just been playing with Adobe Audition 1.5. It makes a very
> fair stab at extracting a vocal - removing the band. There are
> nasty artifacts in the result, but the basis is there.
>
> Any ideas what the process is?
>
> CubaseFAQ www.laurencepayne.co.uk/CubaseFAQ.htm
> "Possibly the world's least impressive web site": George Perfect