I am interested in extracting voices (instruments or vocal) from a song - an idea I got from an old MIT lecture on DSP I saw on YouTube (they're pretty cool and easy to look up so I recommend them). I didn't see anything on the forums. I am a noob with DSP theory. Right now I am working through Andy Farnell's and Miller Puckette's books.

I will be grateful if someone could point me to methods or resources that address this kind of processing. I know there is no quick fix and I will probably have to study on this problem for a while. Just need some help getting started!

If this idea is possible to implement the most helpful response would be if someone could define the problem for me. The article attached is the first result on my Google search and I thought even as a layman I could mine it for a possible solution.

Finally I'm wondering if this processing technique can be done in real time.

Thanks for looking at my post.