Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

ikermalli

macrumors 6502a
Original poster
Hey all,

I want to get a program which edits audio, but I want it to be able to extract the voice from the background music, or the other way around, and I want to be able to put the voice of song A to the background music of sound B, then save it to .mp3 format
 
I'm not sure you are going to find a stock program free or not to do that, as least not easily or completely. What you are looking to do is a very involved process🙁.
 
This is not a trivial process! Extracting voice from background music is a very difficult task. If anyone finds a program that can really do this, I'd pay a lot for it.
 
It's indeed a sophisticated process. Most of the time it's easier to buy the song's karaoke version (iTunes)?

Why do you think karaoke versions are usually re-recorded? Because it's technically almost impossible to do.

Reason: Imagine an oil painting, a tree with red apples on it. The painter clearly painted the green tree first, then the apples. If you remove the apples by cutting the canvas, what you'll get is just holes—not the green tree background.

What you're trying to do is like using solvent to remove just the red paint, but not the green. Somehow possible, but the result will be far from perfect.

BTW, the standard technical procedure is Reverse Phase Cancellation. Eliminates everything mono in the middle of the stereo signal, but seriously degrades overall sound quality.
 
It could, in theory, be done... it would just sound like garbage. The program would need to isolate the exact frequency of what you don't want at every point in the track, and then produce the opposite frequency to cancel it out.

I can't see it working well to begin with, nor can I see an automated program actually being able to figure out what were vocals and what was not... I think it would wind up having way to much trouble figuring out what was what to work. And even if it did, it would be insanely processor-intensive and probably melt your machine...
 
It could, in theory, be done... it would just sound like garbage. The program would need to isolate the exact frequency of what you don't want at every point in the track, and then produce the opposite frequency to cancel it out..

As I pointed out before, this will leave "holes" in the sound. The singer's voice may have covered a softer instrument, say a flute, at a certain point. If you delete the singer's frequency, the flute is still gone.
 
Separating vocal from audio is something that researchers have been trying to do since the ability to do it on the fly would be a great advantage to people with hearing probems. Hearing/understanding in a quiet environment is one thing, but trying to do it in noise/music is a very difficult proposition. Even trying to do it in non-real time is a very messy process and the result usually sounds terrible since music and voice are so close in the audio spectrum. I've spoken to researchers and they say there's really no way to do it without degrading both the audio and the vocal.
 
It can only be done if the sound is repetitive where there is a point in time frame that the singer stop to sing while the misic it self is repeating.
 
Short answer: no, you're not going to get this out of a free app.

Long answer: some of you are saying it is impossible. Actually, it's doable, just hard (read: impossible) to get perfect. I have done it with VirtualDJ and Adobe Audition, and it ends up sounding decent, but not good enough.
 
the best thing to do is to look for an instumental and a acapella track and then mash the two. you can do that in garage band if they are the same tempo

if not then use

tracktion
or
tracktor
 
Many tracks are released as acapella/instrumental (depending on genre), that is the easiest route. Otherwise, you might be able to do it with Audacity and a lot of elbow grease.
 
Hey all,

I want to get a program which edits audio, but I want it to be able to extract the voice from the background music, or the other way around, and I want to be able to put the voice of song A to the background music of sound B, then save it to .mp3 format

Even if you have a bunch of money to spend you are not going to find software to do this for you. It does not exist. But you can get kind of close. There are some tricks, like adjusting the phase of one of the stereo channels (likely by 180 degrees) and adding it back to the other to isolate the voice. (which is likely panned to center) Then you subtract that out. So what you need is (1) some basic audio tools and (2) some considerable skill and knowledge. At least #1 s available for free: http://audacity.sourceforge.net/

Audacity is free and will do what you want. but it will do it in the way that a saw and hammer can build a house.
 
Maybe I could find an acapella of song A then put it on song B, would that work? I am trying to put Kanye West - Heartless on DJ Shadow - Organ Donor and see how it sounds, then mix and match, etc.
 
Oh it looks like I just got beat.

Basically you want to get audacity (not sure if it works on a mac) and import your stereo track into the program.

Then what you need to do is invert the second side (right side). I'm not sure but I think this will take out the vocals (to a degree-it's by perfect and it never will be). There may be another step involved but it's pretty easy to do.
After you do that, make sure you put the second vocals at a high level so you drown out the first ones (there will be a little gargle left)

Hope that helps
 
Organ donor has no words, so is it still needed? Just in the beginning it goes "no wonder your beats are so good, you have no organs!"
 
ooooohhhhhhhhhh you want to take vocals from a song and move them?

yea that's gonna be tough. go with what ChrisA said. It will be tough and hard and you'll probably need to google it, but audacity is capable.

good luck
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.