In a VoIP conversation this causes confusion and people start speaking at the same time. If the latency is big enough the sounds will start mixing. You’’ll find that the computer can’t keep up with the original transmission. This sound exchange is recorded and measured against the original transmission. I think a good test would be to feed the computer sound at intervals (through a simulated microphone) and also let the computer playback sound at intervals. Anything that makes the audio system switch between sources and thus creates potential for measurable and audible latency. I think latency is a keyword because it destroys the experience just like low FPS destroys the experience. A few thoughts on possible audio benchmarks
