Hello fellow music loves,

i was wondering if anyone knows how to remove music from vocals????

Cheers

 

Some say that Audacity can do the job!

 

If the vocals are recorded to one channel or centered, you can try to remove with Mid/Side Eq technique (Side) but not always .

 

1 Mid/Side (Voxengo MSED free) AND low cut to remove kick/bass
2 Any spectal isolate/remover plugin : Izotope RX, Elevayta Extra Boy Pro... More effective, more complex.

 

Rename the *.wav to *.mid and delete the unwanted notes.

 

Are you serious, man??

 

Oh man, that question to begin with...

 

Is there instrumental version of your track?
Both better to in wav flac or any lossless.
Make 2 channels in daw, first for original, second for instrumental one. One should be played with second one, sample to sample. One of them must be played with inverted phase to subtract only music, because waveforms are similar. Vox should remain. If you still hear some quiet music on background, so work with gains,faders, it so because of unmatched levels. Then render and manually restore your vox in izotope rx to clean it from unsubtracted stuff. Why manually? Because only you can find these zones to clean, select them and gain down.

 

Last edited: Jun 11, 2016

Melodyne can help to come closer, but still: you won't get a clean only vocal file - not yet..;-)

 

I was able to isolate a solo flute part from an concerto with orchestra with Sony's Layers Pro. It was manual process, took me 4 weeks to finish but the isolated part was clean. Flute has got easy to see harmonics, not so sure with vocals in a contemporary tune. Give it a go, might work.

 

Find an instrumental track of the song, in lossless format. Find the track with the vocals, in lossless format. Align them so that they are in perfect phase (no comb filtering) and is basically doubling in volume (+6dB). Then flip the phase of either one of the two files.
You are then left with only the vocals.

edit: Oh, I see Milk Pump already gave the same answer

 

ok so this is the youtube vid of it but i cant seem to find an instrumental version of it

thanks for your help everyone by the way

 

ahahah dis is the most noob question any1 can ask.

I know dis because when I first started doing this shit, I actually remember asking a question like this in a now dead music forum...

so cute :D :D :D

 

Find the instrumental. On the same site, find the acapella. Done. Rinse and repeat.

 

Yeah, that would work with dry vocals in mono panned dead center. Good luck finding a track with dry vocals other than Barbershop music.

 

[Confirmed I can't read English anymore.]

Hey, I thought it was making Karaoke not Acapellas...Acapellas is way harder than removing the voice, imo. Even adobe's audition has a specific function to do that and it's quite good. Cause I thought: Okie, I remove the voice with that handy function of Audition, phase invert, add back to original track with voice, viola. Ha, not viola! Eerie, scary, unworldy, and weak.

Otherwise, find a eq that can equalize mid/side and remove it...vocals are almost always in the middle of the stereo field, so, it sits there waiting to be cut down like those birds that just wait for you to walk up to them and shoot them (and ppl say they taste good too...)

I wonder if you really had the time and the patience you could use Cubase's version or Melodyne to just have it analyse the song, and delete the specific blobs. I think that would be the most "perfect" surgery, but idk.

As mercurysoto said, most tracks actually are "released" with the vocal cut out by the artist themselves. That'll take money...and it's no fun, but it is perfect.

Hey, I lost my place in comment line and now I look redundant...that's ok...

@Pipotron3000 oh crap, RX has that function? wtf was I learning, when I was learning their mastering plugin...I thought it was RX 7, I'm confused...

@drakem20 Do you mean you were taking out the formants of all the other instruments? Holy crap that sounds like fun

I've tried phase inversion, and it's still only as good as the instrumental you have...for Acapellas. It's eerie as you have less and less high res stuff.

 

Last edited: Jun 16, 2016

There's actually an amazing open source project some guy made at CCRMA for a thesis that can separate music/voice near perfectly, but it's pretty rough around the edges, slow as hell, and getting a perfect sound is tedious and takes forever. It DOES work though, and it's neat to play with for a bit.

http://isse.sourceforge.net/about.html

 

A pretty damn tough undertaking; regardless of methodology. Probably one of the most asked (audio) questions in the known universe, Melodyne can certainly perform some minor tasks, but ultimately, the final product will be quite a bit less satisfying than anticipated. Maybe one day this endeavor will yield some passable results. Unfortunately, the current available technologies are not there (yet). I'll say one thing more, whoever develops an effective (truly workable) solution will be the toast of the audio world for decades.

 

"Open source" is always a great first step. Get every qualified body and mind involved. Who knows, the technology might be closer than we all think. My hat goes off to all the courageous souls that have been trying to solve this task for as long as I can remember.

 

It's a fun puzzle to try to solve. It seems like it'd be SO easy because it's so simple for humans to identify the different parts. Then you look at the raw sound data and realize how absurdly hard it is. Gives you a new appreciation for the brain. I think the first good, workable solution will come from a neural network. Google's DeepDream network they developed for image recognition is heading down the right path