[Thin Lizzy]

Hi,

What I am about to ask is probably just plain silly.

I will try to explain this the best I can. I want an app that would be able to take the vocals out of a song, and not use the rest of the song. Then, I want to remove the vocals from a different song, and use the music, not the vocals. Then, I want to combine the vocals from the first song, and use it with the music of the other.

An example. I take Michael Jackson's vocals out of the song "Beat It" Then, I remove the vocals out of Van Halen's Pretty Woman, ditch the vocals, and use the music. It will sound like Michael Jackson singing in the Pretty Woman song, with out David Lee Roth singing in it though.

It sounds like something stupid, but I always wanted to try this. Anything software that can do this?

Thanks!

[Thin Lizzy]

Bump!

[Visnaut]

I'm no audio expert, but once a song's been muxed down into a flat audio file, it's nearly impossible, with today's technology, to separate different tracks out. In our mind's eye we can isolate many different sounds and instruments, but all a computer sees is a complex set of sound waves and frequencies with no context. Much like you or I hearing a foreign language with no root or base words.

What the human brain is capable of doing in that regard is quite incredible. Unfortunately, no one's been able to turn this into a working piece of technology.

Karaoke machines sidestep this by filtering out the frequencies that the human voice is capable of producing.

So you're out of luck unless you actually have individually recorded instrument tracks, in which case, you wouldn't be in need of such magical software.

[Randman]

Didn't Audion used to have this feature?

[chabig]

Often the vocals are centered in the stereo field, so if you subtract one track from the other (e.g., R-L) you get a mono track with little vocal left. But if the vocals are not centered, or the audio has been processed, the results will not be very satisfactory.

Chris

[tooki]

For all intents and purposes, what you're asking is impossible.

tooki

[analogika]

Originally posted by chabig:
Often the vocals are centered in the stereo field, so if you subtract one track from the other (e.g., R-L) you get a mono track with little vocal left. But if the vocals are not centered, or the audio has been processed, the results will not be very satisfactory.

This will also remove ANY OTHER signal that is centered, generally including the kick drum and bass.

The only way around this is to separate the signal into three frequency bands, one below vocal range, one exactly encompassing the vocal range, and one above, and to only invert one channel of the mid-range (vocal range).

that requires fine-tuning for even moderately useful results, but AFAIK that is how those karaoke boxes work, including Audion.

This has got increasingly difficult, however, since virtually all recent productions have been completely futzed up in the upper range with exciters and stereo enhancers doing all sorts of awful phase-shifting/addition crap (akin to the atrocious sound "enhancer" built into iTunes).

Note also that partially removing some voice for karaoke purposes is VERY different from extracting voice from music, or indeed entirely removing vocals from a mixed and mastered track. Those things are, as tooki says, quite impossible, and will likely remain so for as long as only finished production mixes (rather than raw tracks) are made available.

-s*

[TETENAL]

I like the iTunes sound enhancer.

[analogika]

Originally posted by TETENAL:
I like the iTunes sound enhancer.

With all due respect:

That simply means that your audio system is crap.

[Thin Lizzy]

Hey guys,

I really appreciate your responses. I was hoping that it would be possible. I could have had some fun!

I dont use iTunes sound enhancer. I dont mess with any of that stuff, except the iTunes Crossfade. I like that. My audio system is pretty cool though. 5 speaker Yamaha Home speaker set, and a powered Kicker 10'' subwoofer. Sounds great!