Well I'm not the expert on this one, thats for sure. Yet from what I understand thats a process that requires the skills of a knowledgable individual with the right sorts of programs. In otherwords, not feasibly doable for the average DIY'ers.
You require a program that can isolate the various aspects of the music from the background noise as you've mentioned, and that task has been made harder by the method in which you recorded it all. That being over a format that compressess a lot of the information, so much of the data that a program could use to do this is lossed or altered.
The real problem in my mind is that everything on that recording fits within the audible frequency range for us. So for instances the voices of people screaming and yelling in the background occuring in the same instances as the singers voice are inseperable without having recorded directly from his mic.
Hopefully some more in the know people, perhaps a sound engineer or two, could shed light to this subject and help you out better.
