Are voice messages 'compressed' after they're read from the soundcard before they're encoded and transmitted?
I'm taking of the kind of compression that is a kind of singnal processing that reduces dynamics, not the one that reduces data size to make it fit through lower bandwith connections.
At the moment voice volume of different players have huge differences. if i try to increase my mic volume to the same volume i hear from some other players, sampledata sounds like it is clipped. maybe some soundcards (those that sound lound and not clipped over voice seem to use them) have a build in effects processor that applies some kind of compression?
If there's compression build in the game, what parameteres does it use? can you please make them user changeable?
If not, can you please implement it? Only need to calculate some kind of average level and multiply a factor to the data, shouldn't be much more than 5 lines of code or so

an explanation of compression can be found at:
http://www.harmony-central.com/Effects/Articles/Compression/