AAC | ADPCM | Convolution | DAFx | FFT | IIR | Mixer | MP3 | MPEG | MPEG-4

> Hi,
> I am working on a project that involves the separation of a stereo
> recording(wav) into independent components using the DUET
> algorithm.However,the algorithm fails when i try to use it on recent
> recordings.I then tried the algorithm on a 1973 Beatles track and the
> algorithm managed to separate the vocals from the instrumentals.I
> would like to know about the different stereo mixing techniques that
> were used  in the past(60's-70's) so that I can understand why the
> algorithm is successful only for this recordings.

Many old recordings featured the vocal panned dead centre with instrument
panned in the stereo spectrum (or recorded in the stereo spectrum).  In this
case the vocals appear equally on both channels, so subtracting left from
right should leave only the instruments, and subtracting this from one
channel should leave just the vocals.

Due to the technology of the time, many recordings (particularly vocals) are
severely band limited, and vocals were often recorded dry (with no FX)

With modern recording techniques it is common for vocals to be treated with
stereo effects, which spread the signal around the two channels making the
above method of isolating them less effective.  It's still pretty common to
pan the vocals dead centre, but you will also have most of the drums and
bass there as well.  It is common to use delays, chorus and reverbs on all
sounds, as well as other effects and this makes extracting individual sounds
particularly tricky.

This is compounded with the better frequency response because most sounds
have at least some element of their sound spread over the spectrum, where it
would have been band limited before.

The Puppeteer
http://sonikmatter.com
	

(You need to be a member of audiodsp -- send a blank email to audiodsp-subscribe@yahoogroups.com )