http://www.htfr.com/more-info/?catno=MR123284

I'm interested in removing the vocals from some tracks - does anyone know if this is a good way of doing it? And how does this thing work - surely just filtering out particular frequencies would also remove parts of the melody which are in the same frequency range as the vocal?

Also can anyone recommend any other (better?) ways ways of making your own instrumentals?

Thanks.

you will get the same effect by putting a sock over the speakers.

if you had several really really really high quality fully parametric eq modules or paragraphics and put them together you might be able to reduce the vocals a bit while preserving everything else- but thats it.

Everything else will just fuck with the melody or whatever else is in the mid range as well.

Lol. Thanks mate; I did wonder if they were just shitty kill switches...

nothing shitty about kill switches.......

as long as you understand what they do.

bottom line is there is no way to remove a vocal from a stereo mix.

Don't even bother. Just keep an eye out for dubs and acapellas

___________________
Digweed is not from this planet.

Ah actually.... there are areas of research into it. Remember that vocals are quasi-periodic where as most other no-vocal sources are fully periodic.

You could use adaptive filtering to do it but you'd probebly need to make a fully custom (and customiseable) DSP processor on an FPGA (a kind of reprogramable logic, kinda). Which I don't think is worth it really

Maybe someone has done it already (probebly what this is anattempt at actually)?

If it wasn't running at real time I'd think there would be ways to do it (like on a PC, supose even then the adaptive filter algo would still need to be pretty adjustable and take ages to configure), but it would not be easy and not turn-key in that you'd need to piss about with it alot.

At first glance I think this is a bit simpler than the method I was thinking of:

quote:

How can a CD player remove vocals from any song on any CD? By reversing the phase between the left and right channels and adding a complicated DSP equation, the center channel (where most lead vocals reside) is greatly reduced audibly. On some songs, the vocals reduce 100% and on other songs the vocals only reduce a little or not at all. Why is this? The varied results of the vocal reduction feature of the PlayMate Vocalist is totally dependent upon how the CD you are playing was recorded. CDs where the lead vocal is recorded in the center channel, usually have the greatest results. CDs where the lead vocal is NOT in the center channel have less effective results. What is the Alesis Vocal Zapper? The Vocal Zapper works on the same principle of the PlayMate Vocalist, but does not include a CD Player. The Vocal Zapper has 16 different vocal reduction programs allowing the user a variety of choices when experimenting with different CD’s. The Vocal Zapper is contained in an ultra small, rack mountable package. More information on the Alesis Vocal Zapper can be found at www.alesis.com/products/VocalZapper/about .html

>LINK<

But I supose the "complicated DSP equation" could be an LMS adaptive algo but I very much doubt it.

This might reduce the vocals a fair bit but I'd recon you'd still hear them and on some songs it would do fuck all.

Even on a computer (which I have) with some of the best eq's and filters it still won't happen. And I don't see it happening ever to the point of having the song sound as good without the vocals. Simply because every producer sets their vocals diferently for each tune, with complicated reverbs, echo's, and ping pong delays. Even with the simpelest of songs it would still be very hard to do because the human voice has so many different harmonics going on throughout the song. I wish it could be done, but until you find that magic gene bottle I don't see it happening.

___________________
Digweed is not from this planet.

quote:

by Dervish But I supose the "complicated DSP equation" could be an LMS adaptive algo but I very much doubt it.

Could be something like that... but for the price I'd suspect its just a few preset filters

quote:

by hooj1 Simply because every producer sets their vocals diferently for each tune, with complicated reverbs, echo's, and ping pong delays.

Yeah, the effects are the real killers here. These alone already make voices not centered. + the other mentioneds

___________________
http://soundcloud.com/ed_m

quote:

Originally posted by hooj1 Even on a computer (which I have) with some of the best eq's and filters it still won't happen. And I don't see it happening ever to the point of having the song sound as good without the vocals. Simply because every producer sets their vocals diferently for each tune, with complicated reverbs, echo's, and ping pong delays. Even with the simpelest of songs it would still be very hard to do because the human voice has so many different harmonics going on throughout the song. I wish it could be done, but until you find that magic gene bottle I don't see it happening.

The type of thing I'm talking about isn't a static filter it is adaptive to what is happening. They can be used for example to seperate noise and vocals from a line. Even noise which is present at the same harmonics as the vocals.

Like you say each track is different, but a program could be developed which you could use to do it I'm sure. This isn't a filter you turn on and leave it adapts constantly.

Still wouldn't "remove" the vocals just attentuate them, but doing it properly would yield better results. This seems a half assed attempt.

quote:

Originally posted by Freak nothing shitty about kill switches.......

Hehe I know; I just meant that this is a shitty attempt at a "vocal kill switch"

quote:

Originally posted by Dervish You could use adaptive filtering to do it but you'd probebly need to make a fully custom (and customiseable) DSP processor on an FPGA (a kind of reprogramable logic, kinda).

Riiight, so... no easy ways, then?