Some questions regarding noise in voice capture and VoIP

[Tarek Khojah]

Why do VoIP apps use specific codecs? Aren't you sending bytes in a standard format (.wav)? I doubt the audio captured is being converted THEN sent, though I have recently learned that encoding doesn't take as much time as I thought, is it feasible in real-time applications?

Another question, I wrote a voice-chat program that currently works fine over a LAN, but I have some currents issues, and definite future issues:
1-Noise:
The mic seems to capture everything, even air flow.
I am totally clueless about voice (and sound in general), of course I'm willing to read and learn, but where should I start? Shouldn't I set some sort of "threshold" for voice capture? How would I go about that? Just looking for general guidance here.

2-Jitter:
Well, jitter and other similar performance problem that comes with using UDP in a real-time application, is there a standard algorithm for queuing up the packets, re-arranging them, using time-stamps, knowing which to discard (for being "too late") and so on?

Any other recommended readings or general advice concerning the subject (whether LAN voice-chat) or the more commercial sense of "VoIP" well be appreciated.