Extract a vocal from a song AND KEEP THE VOCAL


ehrm, I’m not pretty sure I should post my issue here, and if you think it’s not the right place, please do delete it and tell me where to post this. Anyways, I’m looking for the vocal of Dizzee Rascals “Flex”. I cannot find it on the internet. So I decided to extract it from the song. And as Audacity has proven to be very useful, I ought maybe someone could explain me how to do this?

Thanks in advance,

Signed Pilopew

99 times out of 100 you’re not going to have much luck. It’s like trying to take the chocolate out of a cake.

That said, we have a Wiki page on this topic:

If you can get the vocals removed from a song, then there’s a chance you may be able to subtract the (now vocal-free) music from the original track and be left with the vocals only (more or less). But this is also up in the air.

Once you have the music only track you can do this:

  1. Import both the Music and the Original into Audacity as 2 separate clips.
  2. Line the two tracks up as exactly as possible using the Time-Shift Tool.
  3. Highlight either track (doesn’t matter which) and click effects → invert.
  4. Highlight both tracks and click Project → Quick Mix (or Tracks → Mix and Render).

Seriously though, don’t spend more than an hour trying to get this to work, in many cases it’s simply impossible.

We’ll catch that Wiki up one of these days.

Center Pan Remover (Voice Remover–Search the page for “Center Pan Remover”)

Voice Isolator (Opposite of Center Pan Remover)
Voice Trap
Extra Boy

And all of that will get you into the 10% to 20% success area. These tools are painful production slaves. If the singer isn’t in the middle, they fail, if the singer has stereo echo added, you’re dead, etc.

Most of these tools are location-centric. If the singer and the drums are in the middle, you can’t separate them.


I tried this process with ‘A Forest’ by The Cure and all I did was extract the vocals ( pretty good at, say, 98% removal) and put them back again using the above instructions. So, if I was able to extract the vocal, why was I not able to remove the rest of the track when I compared track with no vocal and track with vocal?

I don’t understand what exactly ‘invert’ does, but Im wondering if there is no process for removing the stereo pans as opposed to the common audio between two separate files, ie the middle audio.

but you did warn me it might be a waste of time… :cry:

“Invert” does exactly what it says - it turns the waveform upside down (inverts it)
Everything that is on the plus side of the central line becomes negative, and everything on the lower (negative ) side of the line becomes positive.

Here are two mono tracks. The bottom one is an exact copy of the top track, but it has been “inverted”.
If you mix these two tracks together, the result will be silence - a flat line - they will cancel each other out.

Removing the vocal (simplified):
On the Left channel you have “instruments on the left” + “vocals in the middle”
On the Right channel you have “instruments on the right” + “vocals in the middle”
(any sounds that are in the middle of the stereo mix will come equally from both left and right, giving the impression that they are in the middle)

“subtracting” the right channel from the left you will get:
(“Instruments on the Left” + “Vocals in the Middle”) minus (“Instruments on the Right” + “Vocals in the Middle”)
as an equation this could be written: (IoL + ViM)-(IoR + ViM)
What you are left with is “instruments on the left” minus “instruments on the right”
(the left channel and right channel vocals have cancelled each other out.)
So the result would be the “left channel instruments” and the inverted “right channel instruments”.

Because “instruments on the left” and “instruments on the right” are different, there is no way to cancel them out.

Note 1: The above explanation is much simplified, but will hopefully give you the idea of how it works.
Note 2: The plug-ins suggested in the earlier posts use other more complicated methods to attempt to isolate vocals and remove all other sounds.

Steve, thank you for your time in explaining this so well. I understand this now. I did try the plugin (VoxReducer) but it didnt work at all. Maybe I should go back and try again.

Thanks :wink:

“voxReducer Kit is designed to reduce the level of the main vocal track in a stereo music audio file”
This is a commercial program that costs $24.95USD,
why not use the free “Center Pan Remover” plug-in for Audacity.
See here for details: http://audacityteam.org/wiki/index.php?title=Vocal_Removal

I discovered a way to get a reasonably good isolated vocals. First, using the method previously described[1] remove the vocals from a song. Using the song with vocals removed, sample it as noise profile with the noise removal effect. Apply noise removal to the original audio.

With a few adjustments, the results can be simple to achieve and impressively good.

[1] http://wiki.audacityteam.org/index.php?title=Vocal_Removal

Wow, I am really impressed. That’s a much better method than anything I’ve ever heard before. It’s still tricky to get rid of anything else in the center (this almost always includes the bass guitar and bass drum), but careful use of EQ can get you a pretty good result.

Thank you, Marco.