No, the usual way that these things work is to invert the phase of the R channel and add it to the left channel. This cancels out anything that is exactly equal in both channels, i.e. centred in the original stereo image.

There are many problems with this approach. For example, bass guitars and kick drums are normally centred, so they will be attenuated as much as the vocal. Additionally, reverb that is added to any track before mixing is usually stereo (has different L and R components) so does not cancel, with the result that even if the direct vocal is suppressed, you still get much of the vocal reverb.

Whether the commercial VSTs work for you will depend entirely on the material you give them to work on. Some may appear to perform an acceptable job. Others will just sound wrong.

Ahh... I see.

Thanks for that Boswell. Now I understand how they work, Im going to have a go at replicating the technique.

I take this one step further to lessen the blow. The bulk of the range of the human voice lies between 300-3000 Hz. So what you do is you separate the stereo track because you're going to process one track differently than the other. On one channel, you will not equalize anything. No playing with tone controls. On the other channel, you will utilize a software filter to roll off everything below 300 Hz. Then when you combine the left and right channels out of phase the low frequencies will be uncommon enough to each other to remain. It is out of the range of the human voice and the voice will cancel. This keeps more of the original feel and flavor intact. And nobody's preset does what I do. Nobody's software does what I do. I've never been able to figure out why everybody else hasn't figured this out yet? This can even be done in a stereo widening method though it's a lot trickier.

It's fun!
Mx. Remy Ann David