This is far more processor intensive than you realize. For example if you have five players in a game, each in different areas with unique materials and sound reflections (eg: upstairs, ground floor, basement, backyard, and front yard) that are all shouting out to one another, then you have to apply precise filter and reverb/echo effects to twenty audio streams in realtime. This is achievable but the processor cost can quickly get out of hand. And how do you deal with audio input lag?