Is there much progress out there in audio processing algorithms to remove lead vocal from music tracks to produce a karaoke, or the inverse, stripping out everything but the lead vocal to produce an acapella?

There is a plugin for WinAmp that will do that for you. It only "sort of" works.

There are other vocal eliminator products that do similar things, and they also only "sort of" work.

They operate on the principle that the main vocal track is usually recorded in mono and is perfectly centered in the stereo field, while the instruments are usually panned somewhat to the left or right. They work by subtracting all sound that is mono/centered.

This only works on some songs, and the remaining leftover music is often thin-sounding and missing some instruments.

Another problem is that there is sometimes stereo echo effects applied to the voice, and that echo remains in the track after subtracting the vocal.

It's *very* hard to do. There are plugins, filters, etc. for WinAMP, VST and DirectX that make an attempt to do this, but they end up stripping a lot of the real music out of it, too, and it doesn't work for all songs anyway. It depends on the vocals being mixed in a way that they're expected to be, and although a lot of modern recordings follow traditional mixing rules, some don't.

Do a google search for "vocal eliminator" to find some companies that will do it for you, or if you wanna "try it at home" check this guy out.

Edit: Damn the FAQbot got me again. I'm going to just stop posting here and move onto a BBS where the users are all humans.

Hey, at least you linked something specific for him to try.