On Xbox360 XHV2 is used for dealing with voice data, and XAudio2 is used for the low-level processing of that data. What is the XHV2 equivalent under Windows, assuming FMOD is used for the low-level audio processing? Previously DirectPlay.Voice was used to facilitate this.
There isn't any standard feature built into Windows for this that I know of. There are libraries that will do it for you like RakVoice for example, although that's integrated with RakNet. I believe the Steam APIs also have it built in.
Unfortunately Fmod lists it's voice chat support as coming in a future version - see http://www.fmod.org/index.php/products/fmodexdetailed and click on "Enhanced Internet Features".
That means you'll probably have to do it manually:
Fmod obviously has APIs for steps 1 and 6. It also has audio codec support so it should help a bit with step 3 and 5. The main thing you need to do is step 4.