Hi… Just recently registered my copy of Renoise and I’m so very happy about it : )
Anyway, this isn’t a Renoise specific question, but how does one sample or “extract” the vocals from a song? That is, when various people do remixes of pop songs etc. how do they isolate the vocals so they can use them in their own tracks? Xerxes seems to be very good at this : )
If this has been addressed already please point me to the post.
Thanks!!
yh
Well if you can kill the vocals, you can go another step and simply invert the vocal-less waveform, and add it to the original one.
Or so I think… 
edit: Just saw that the trick was for hardware output… so nevermind.
And that’s actually pretty much the way simple ‘prologic’ surround works (which was originally Halfer Surround).
And if you wanna do the same with our favourite software called Renoise, do following:
Load the desired sample in Renoise, go to Track DSP and add 4 or 5 stereo expanders to the track. Set all expanders except the last to max. Set the last expander to mono. Now you’ve got the 3rd pseudo-channel out of the matrix.
For the other way round (keep voice, remove background), simply set the ‘Width’ on the source DSP (TrackVolPan) to max.
The magic behind it all is channel inverting as the helpful guest and teknocide already said. Damn, you guys were faster! 
wow, ive asked this loads and was alwys told u couldnt! will give it a try!
ok, um so it doesn’t work.
Stromboy, which way round? Keeping vocals or killing vocals?
As it’s mentioned already, keeping vocals is harder because there are too many mono sounds which come along together with the singing.
Killing the vocals works indeed, actually. I can send you an rns which shows an example of both if you want.
Could you?
Ive not had much success with either unless im choosing difficult tracks! 
That’s two patterns. One is supposed to remove the background (which doesn’t succeed too good) and the other one removes the voice.
edit: checked already? did it help?
If your track is already quite noisy, grungy or whatever, and you don’t mind the vocals having some dirty-ing up applied to it, by applying some filtering (esp. bandpass/highpass) you might find that what is left of the background “noise” gets drowned in the rest of your tune anyway, or even adds a certain “je ne sais quoi” 
If you have a soapy clean tune, you’re less lucky of course
I completely forgot about this one:
There’s another big trick you can do. YoutH reminded me of it:
Either use Cool Edit (Adobe Audition) or Sound Forge (with the Noise Reduction PlugIn) for it:
The intelligent noise reduction works like this:
You mark a part (usually noise) of your sound. NR gets this as a noise profile.
Then these analyzed frequencies will be removed from the entire sample.
You can also set the intensity and noise bias(and some more) in the usual Noise Reduction.
Now you might say: Yes, but it’s supposed to remove noise, why should it be useful for me?
Well, if you have some background-sounds left in a solo-part in your sample, and if they are the same which come together with the vocals, then you can use the Noise Reduction to remove the background from the voice.
you can try phase cancellation, I havent tried it as of yet.
This is a post from another board by a friend of mine that has been working on this :
"as far as i know, people who have discussed this topic were only talking about it in theory, if you have the original tune, and the instrumental mix, you can invert the wave file of the instrumental mix, and mix them together to cancel out those frequencies, leaving everything else.
after many only semi-succesful attempts, i found out what the key thing to do is.MONO both tracks.i suppose you could either take one side or the other, but i chose to mix/expand (mix both channels, but leave them seperated) then you have to take the EXACT same section form both OG and Instr. versions, copy the inverted file to the left channel for example,and then mix/expand again.as it’s not usually perfectly exact, it seems to leave a tiny bit of the beat in, but you can usually eq it out.
technically it IS possible to get it perfect. i found that out by testing it with the exact same sample. it erased everything!!!
i’m very pleasently suprised that this actually works. now i don’t have to bitch about not getting a sample i want.
to note:
mono’ing helps cut down on the slight differences in the wave file, because the more they are off by, the more will be left over. i still haven’t spent ALOT of time doing this, but the test’s i’ve run look really good so far. with something like this, it pays to be anal, and spend as MUCH time as possible, to get it perfect. "