Reasonfreaks

[tranceformation]

I'd like a phase inverter or flipper for making accapellas - extracting vocals out of mp3's, wav files etc.

[celestialmind]

Phase inverters are already there, google dude.
as far as extracting vocals out of mp3's or wave files you can forget about it 'cause the only way to get an actual accapella is to contact the producer of the track who has the vocal on a separate channel, as a separate file and get it like that. mp3's, waves, aiff's, whatever file extension you want are ENTIRE files, it'll never happen, there is no such thing and never will be, I can guarantee you that. Don't try to eq it out either yourself 'cause you're then unknowingly deducting low, high frquencies and overall dynamics of an actual vocal, so.... yeah.

[bastard_son]

I don't think reason will ever be able to extract vocals, since it's a pretty heavy operation to do on a soundclip. It's not exactly related to making music. It's more like a reverse engineering procedure.

Like CelestialMind points out, it's not possible to extract a perfect vocal line from a song, but there are people out there trying to get as close as possible.

One of the tricks they use to avoid cutting the frequencies are to invert the phase of the music. This of course requires the song to have a part where the music without vocals is identical to the parts with vocals. You can also do something similar with a profile noise reduction effect.

Another trick is to separate the stereo field, because the vocal usually has it's own space. I remember reading about it, but I never tried it, and honestly; I have no idea where to begin on that. You need some good scientific tools to do that.

[celestialmind]

Izotope ozone can do multiband stereo imaging, HOWEVER, you need to be aware of what's happening to an actual vocal:
each vocal is gonna have it's own section of dynamic frequencies, there will be some words like "duh" that have that little bass to them, words like "systematic" that has the sss's and high frequencies to it, and so on and so on. Stereo imagine may seem like it's helping but it's subtracting actual frequencies and the dynamics. Stereo imagining, unless practiced and used a lot, can be very destructive in general if not used right.

Bottom line, you cannot extract vocals out of actual tracks and never will be, it has been wished upon and talked about million of times, sorry, no can do.

[tranceformation]

Bottom line, you cannot extract vocals out of actual tracks and never will be, it has been wished upon and talked about million of times, sorry, no can do.[/quote]

Errrr,sorry but it can because people have done it.
Visit www.acapellas4u.co.uk and read the DIY Accapellas section. There's a whole host of extraction methods in there tried and tested, including Phase Inversion. They suggest Cubase as a suitable software for doing it.

[theoracle]

it's still not perfect though, but with enough cleverness up your sleve you can mask the hint of music or noise behind those vox with your own music.

[bastard_son]

yes good point. Given that the vocal is somehow harmonically related to the music that you use it in, it should blend nicely since the little leftover noise/music is in the same frequencies as the vocal. It's not that you have to follow the original chord structure or anything, but at least pay attention to the notes in the melody line and make sure that it works with your track..


@CelestialMind: I'm not sure I understand your post correctly; is Ozone capable of splitting a stereofield into segments? I don't mean adding a separated stereofield to a sound, but rather subtracting it from a mix?

Something like:

[tranceformation]

I'm going to be remixing two famous songs together in one song soon. I'm using the Lyrics from song "A" and playing them over song "B" by using the NN19 Samplers in Reason 4. You'll be amazed at how they fit together so nicely although I'll have to tweak the timing of the vocals slightly because the pieces are at different speeds and pitches but they are only a tone apart so the vocals won't go through too much of a dramatic change in terms of pitch. I'll let you know when I've done it.

[butchmanring]

Better to pitch the music to the vocals using an algorithm that can preserve the formants.

[tranceformation]

Hey, my remix sounds great. Well, my brother says it's known as a mash up but it's basically the same idea. The two pieces go together like Jelly and Custard. I might post it on a website so long as I can be sure no one will steal the idea!

[celestialmind]

Bastard Son, there is 3 major things about what we were talking about earlier (stereo field):
1 - you can adjust what frequencies you want to alter with what band
2 - you can bring together or spread apart each band
3 - you can phase and delay the stereo imaging for each band

I hope that helps.

[shauk]

http://virtualturntable.fourstones.net/ ... n-tutorial


http://virtualturntable.fourstones.net/ ... ng-vocals/


there are ways to do it. It's just incredibly time consuming.

[about2flip]

I know this is an old post, but I am putting up for anybody reading this and needs this. Use Music Morpher Gold. It has a voice extractor.
www.musicmorpher.com/free-tutorials/voice-extractor.htm

or google extract voice from music.

It should be the 1st or second link.