From what I’ve read around the net, completely removing music and leaving just speech is nigh on impossible.
What I was wondering is if there was a decent way, Audacity or otherwise, to just quieten the music if not remove it, while keeping the vocals at the same level?
Technology is our friend. I posted an explanation last night just as the Audacity web site crashed. ~!@#$%^
There has to be a unique key that lets the software figure out what’s voice and what’s everything else. In the case of Vocal Removal, the key is any sound in the exact middle of a stereo performance. Drums are frequently in the middle and they go too. If you violate any of those rules it stops working. Singer not in the middle, sorry. Mono original show, nope. Stereo echo added to the voice, not happening.
The illustration for how bad this is comes from two mono shows. One an alto Saxophone and a tenor singer and the other, a three-piece vocal performance backed up with the Vox Humana stop on a pipe organ. In both of those cases, the “orchestral” portion is indistinguishable from the vocal performance. There’s no key, and in those two cases, it’s not unusual that humans listening can’t tell who’s playing.
People will insist you can use the various bandpass filters to isolate the human voice from everything else. They say that until they try it. Then they stop saying that. The human voice has a really wide range and rich subtlety and is extremely difficult to isolate without destroying it or just not working at all.
If you get something to work, post back.