I have next to no knowlege of the inner workings of mp3 compression, but I thought I'd ask this in case I happen to be the only one to have thought about it:

Would it be possible to improve the bitrates of live performances by starving the crowd noise of bits while retaining higher bit rates for the actual music? I don't know how it could be done except by some sort of frequency analysis, but I imagine any sort of noise would be hard to encode, so not only would it be taking up more bits than the rest of the signal, but it would also be something that people aren't particularly bothered about hearing in perfect sound...

As I say, it's just a thought - I hope it provides some sort of use and apologise if it doesn't .

apart from the technical side... why would you degrade the quality of the background crowd? after all this is what live events are all about. the singing, shouts, simply the whole feeling.
for most (mainstream and I guess most others too) releases there are studio versions available.

You would have to seperate crowd noise from noisy instruments. Hard. Note that noise is broadband so you can't just cut some frequencies.

I don't know too much about it but I would think that isolating the crowd noise would be quite hard.

A better way would be to give a big bribe to the soundboard guy so you can get a direct patch. That way you won't get much crowd noise to worry about. Plus that way you don't have to deal with the drunk guy next to you who keeps shouting for the same song all night long.

I appreciate people want to hear the crowd's noises to some extent, but I think a lot of the general "noises", not the individual shouts could lose a lot of bitrate without much audible loss. Anyway, a few people have said it's far too difficult to be of any use so it's a moot point, really.