Separate beat and vocals

[mrk_webber]

Hi, glad to finally find the Logic Pro forum. Need your professional knowledge and help – is there a way in Logic Pro 7 to separate the beat and vocals from a original file coming from a original CD disk. Or is there a program who is able to do this?

I want to use some beats from tracks from a cd but I cannot get rid of the vocals, any idea’s?

Thanks and look forward to hearing from you.

Mark Webber

[northernlogic]

There really isn't any way to accomplish this without significantly altering the sound of the beat. Once the tracks are mixed, they're mixed. There have been attempts to remove vocals by reversing the phase of either the left or right channel, but this cancels everything that is panned to the center- including the bass, snare drum and kick drum- not exactly the desired effect. You could try using a parametric EQ to carve out some of the voice, but you'll lose a lot of the midrange, where the snare and toms are most prominent.

One thing to try- if you are working with a repetitive beat, look for breif moments in the song when it is not obscured by the vocal. You might be able to find enough "free" sections to splice together into a new, composite beat.

[David]

I've had some decent results by mixing the two methods described by northernlogic: take the Left channel, eq the kick and bass out of it as much as possible, and then reverse its phase. You'll loose some snare, but the vocals are pretty much gone (you usually still have some of the vocals' reverb) - pretty impressive.

[northernlogic]

Neat. It never dawned on me to EQ the bass and kick out before the phase reversal. Of course, I never use prefab loops or sample from other people's songs and I rarely work with vocals, so that might explain why...

[Lz Music]

I've had some decent results by mixing the two methods described by northernlogic: take the Left channel, eq the kick and bass out of it as much as possible, and then reverse its phase. You'll loose some snare, but the vocals are pretty much gone (you usually still have some of the vocals' reverb) - pretty impressive.

David, could you explain "reverse its phase" ?, how do i do that?

Thanks

[David]

Hey Lz... btw, great soccer action! Wow....

Reverse the phase of a signal means invert its polarity. If you look at the waveform, that means the positive becomes negative, and vice versa. If you look at your speaker membrane, it's going to move backward instead of forward.

So if you add two identical but out-of phase signals, they are going to cancel each other. Now the trick in this method is that we add Left and Right, which are not totally identical. In fact, the only portion of L that is identical to R is whatever the mixer has panned right in the middle. Typically, that means Kick, Bass and Vocals.

If you filter out the Bass and Kick from the out of phase signal, you'll avoid cancelling those. You will only cancel the vocals, and whatever is panned in the middle (That often means the snare and some other instruments).

To invert the phase of an Audio Track in Logic insert a Gain Plug-In (in "Helper") and click on "Phase Invert".

[Lz Music]

David, yeah, that's a great move, called the rubberband ... thanks for replying, i'm gonna try that right now to see if i get some good results

Thanks again,

[camillo jr]

I'm trying this right now but I've still got plenty of vocal left, and not just in the verb. What I've done is copied the region to another track, made them both mono, panned each track left and right and then inserted an EQ and a Gain plug on the left channel. I used the Logic EQ's low cut filter set at 160 and then reversed the phase on that track. The song I'm using is Sting's Feilds of Gold, which just last night I did a Karaoke version for a wedding. Interesting notion that I might have saved myself some work this way but it was fun to see how close to the original I could get.

Anyway, what am I doing wrong with my left and right mono scenario? PDC is on.

[David]

I'm trying this right now but I've still got plenty of vocal left, and not just in the verb. What I've done is copied the region to another track, made them both mono

That's not what I suggested. You shouldn't make the tracks mono, you should make them Left only and Right only. (click-hold on the format button to access those settings). And don't pan those tracks either, keep them both centered, otherwise it doesn't work.

[David]

OK here is an example.


ORIGINAL
http://logicprohelp.com/grammy/Vocals.mp3

LEAD VOCAL STRIPPED
http://logicprohelp.com/grammy/NoVocals.mp3

[camillo jr]

Wow. That is effective!

[camillo jr]

Alright, just got a chance to try this. That works really well for removing the dry part of the voice although in my test song Sting's verb-drenched voice is still floating around back there, as to be expected. Great technique!

[CuzCutz]

You shouldn't make the tracks mono, you should make them Left only and Right only. (click-hold on the format button to access those settings).

Where is the format button??

Thanks

[David]

Where is the format button??

Below the meter on the channel strip.

[CuzCutz]

Where is the format button??

Below the meter on the channel strip.

OH... That button.... Thanks

[CuzCutz]

David,

I cant seem to make this work...

I made 2 tracks of the same source material, Made one track right & one track left
On the left track, inserted a ch.EQ and the gain plug
with the EQ filter out as much bass and kick as possible
and with the gain plug just invert the phase.

What happens next, do i just mix the tracks together, Because that does not seem to work.

what am I missing?

thanks

[Ed Ryan]

If you can strip the vocal...

Does that mean if you bounce the vocaless version and then invert it against the original...you can then strip the music and keep the vocal ?

I've tried this sort of stuff before but not had much luck

[David]

Does that mean if you bounce the vocaless version and then invert it against the original...you can then strip the music and keep the vocal ?

Yes, but your results will be less than perfect. With the technique I used, what you're really doing is isolating all the material that is panned dead-center, above a certain frequency range. That's vocals + quite a few things.

Now in the first situation, you take that material away. So you take the vocals away, but also probably the snare, maybe some guitar, etc... usually you'll notice the track sounds different than the original, but it's not obvious what happened. You may have lost the snare, but all of the reverb of the snare is still here, all the bass frequencies from the snare are still here. Same is true for guitars or anything panned center in your mix. The only obvious thing is that the vocals are gone. That makes the technique pretty efficient.

In the situation you discuss, you're left with only that material. Now it becomes very obvious that you don't just have vocals, but also a lot of the other material.

Once again, depending on your original track, your mileage will vary.

[robertryda]

Hello! I've spent some time before i found this topic on taking out the acapella out from a track by using the exact phase-inverted instrumental. This was always a success. But now cant achieve the same result by using these described techniques, no matter what i do. I really tried hard and even spoke to all my musician friends that could help and its all helpless.

I create one stereo audio track in Apple Logic 8, and put a stereo audio file in it (instrumental+acapella all together, a regular song), setting the track properties as "left" channel as described above. Then I create a copy of this track and copy my stereo audio file (song) into it, changing its properties to "right" channel. And so I have two exact separate stereo audio tracks, each representing its own "left" and "right", sitting there for me.

Next I take one of the two channels and put a "gain"(stereo) plugin on it in the second position (there are two - mono and stereo versions of the plug, I believe it has to be stereo as far as my original song is) and invert its phase. As a result I receive MONOPHONIC VOCALS coming out (i beleive those are vocals that were panned off the center in the original song, like backgrounds and reverbs) with little artifacts (dirt) in it. Then I use two exactly set equalizers for the both tracks to cut off low frequencies, which results in giving me less artifacts on that (put them before the gain plug, meaning in position one).

Now no matter what I do i cant get the center vocals cut out of my original song by using the acapella I just got from doing these steps. Well its still the same voice and its supposed to remove AT LEAST like.. 50% of itself or something. I mean.. I guess it has the same frequencies, levels in it, so it should work. Somehow it doesnt. I noticed that different tracks might differ in results. I even tried to work with the audio that David has provided, but wasn't even close to what he achieved.. My brain is exploding and I cant help it. Please help me with an advice on what i might be doing wrong here.. The idea is still to get the INSTRUMENTAL out of the full song and NOT the acapella or whatever comes out in my case.

BIG THANKS!

P.S: I might create a project file and attach it here to make it more clear on how i go through the process, if needed.

[David]

Robert, you didn't follow the instructions to the letter:

so I have two exact separate stereo audio tracks, each representing its own "left" and "right", sitting there for me.

You have two separate MONO tracks. Each one is playing back ONE side of the stereo audio file.

Next I take one of the two channels and put a "gain"(stereo) plugin on it in the second position (there are two - mono and stereo versions of the plug, I believe it has to be stereo as far as my original song is) and invert its phase.

Since the channel strip is now mono, you should use a mono plug-in.

Then I use two exactly set equalizers for the both tracks to cut off low frequencies

You should only filter ONE of the two channel strips, not both.