JR-
> > Anybody please tell me by what means, a voice can
be removed from an
> > audio signal.
> >
> > Is this just an application of band pass filter on the actual signal
> > or completely different approach?
>
> I would imagine you could remove some aspects of the voice signal
> through band reject filters. By far the best way, if you have access to
> the vocal track or acapella is to to phase invert the vocal, and then
> mix with the original. This results in a completely clean and full range
> instrumental track - but this method is only useful if you have the
> vocal track.
>
> Maybe some kind of formant reject? Sorry if that was useless!
If you don't have the original vocal signal, then filtering is mostly
useless --
frequency overlap between lead vocal and music content will be in many cases
indistinguishable. One common technique in the music / performing arts business
is
to work with stereo recordings -- the assumption is the lead vocal was recorded
in
the middle, most instrumentals and backup vocals are on one side or the other,
so L/R
channels can be subtracted to reduce the lead vocal. You might try searching
Google
for 'lead vocal reduction'.
I think some of the popular music editing software, such as CoolEdit, GoldWave,
and
YoGen, include vocal reduction algorithms based on this method. I've not
seen
inexpensive off-the-shelf that purports to reduce lead vocal on a mono
recording.
Such an algorithm would be adaptive and substantially more complex. For that
you
might search professional products, such as Superscope Technologies.
-Jeff