how does one? rip an acappella from a track

  xy_politics said:

Just to add, below is an attachment from another post - Tuss / Synthacon 9, with the vocoder lyrics extracted, using AA2. And that's off of a really bad quality MP3, so you might get away with it.

 

linky

 

edit: this thread was started a year ago... way to go hyperbotfuk

yeah i'm quite aware but i was asking an additional question.

  xy_politics said:

Adobe Audition 2 has an excellent vocal extraction tool - after you phase invert your instrumental against the original, which may or may not work (you will have to be totally accurate lining them up, 1 sample off either way and it will fail, and your relative levels will have to be perfect too), try using it to clean up the remaining noise.

thanks, I'll try this.

  xy_politics said:

If you're using MP3s to do this, I'd say don't bother.

Is this just the standard audiophile "OMFG MP3Z SUX0R" talk or is there a specific reason for this?

  reluctant ebola washer said:

how does one? rip an acappella from a track, when you HAVE the instrumental version

 

 

That's the dumbest quesiton i've heard all day, and I'm a tech support agent!

yet another dickhead

  HYPERFUKBOT said:

  xy_politics said:

If you're using MP3s to do this, I'd say don't bother.

Is this just the standard audiophile "OMFG MP3Z SUX0R" talk or is there a specific reason for this?

nah, to be able to perfectly remove the track from the vocals, you will need as perfect copy of the original as possible. the way that an mp3 gets rid of data will mean that on a sample by sample level the rest-of-track data on an instrumental version compared to a vocal version will be pretty different. a wav is 'true' quality so the parts of the wave that make up the instruments in a track compared to the vocal version should be pretty identical (if they were rendered that way) so you should be able to get a near perfect naked vocal.

however, you may be lucky with mp3's, so its worth a try.

  welcome to the machine said:

  HYPERFUKBOT said:

  xy_politics said:

If you're using MP3s to do this, I'd say don't bother.

Is this just the standard audiophile "OMFG MP3Z SUX0R" talk or is there a specific reason for this?

 

nah, to be able to perfectly remove the track from the vocals, you will need as perfect copy of the original as possible. the way that an mp3 gets rid of data will mean that on a sample by sample level the rest-of-track data on an instrumental version compared to a vocal version will be pretty different. a wav is 'true' quality so the parts of the wave that make up the instruments in a track compared to the vocal version should be pretty identical (if they were rendered that way) so you should be able to get a near perfect naked vocal.

 

however, you may be lucky with mp3's, so its worth a try.

oh yeah, you're completely right. i didn't think of that. but that's only if you try to do the whole invert+add thing and not some more fuxible frequency shit

Also, in many cases the stereo imaging of a track is narrowed for MP3 compression - the main factor in extracting an element from a track cleanly is its unique position in the stereo field, so this can cause hassles.