> I don't think you can really do this accurately enough to get good > sound, but the basic mechanism is time.sleep(t) which takes a floating > point argument. That turns into the appropriate microsleep, I think. I think the time would have to come from a hardware clock.