There a few tools out there, both commercial and open source that may be used to split a piece of music into individual tracks, subject to some limits.
These implement some sort of AI models to identify the individual elements of the piece, for instance vocals, drums, bass, piano.
One of these is Spleeter by Deezer, a research project from Deezer, which is available to use from command line or as a Python Library.
There are a few projects out there which implement the library, the one I've found and like is SpleetSpace, which is great in that they've packaged it well so that it can be downloaded and run with no additional steps or fuss
Reference: GitHub - SpleetSpace
The only issue so far is that the download from YouTube isn't working for me, I suspect as a result of changes at the YouTube end, but that's a nice to have really
The results are fascinating as much as useful - being able to pull out just the bass from a song for example, where it's barely audible in the full production.
Obviously copyright conditions apply!