Actual Realtime Inference Using a Microphone/LLM Audio

[fire17]

Hi there
really interested in the realtime capabilites, but could not a guide on how to run this in realtime

would be awesome to see an online demo with a microphone to test the realtime aspect
and the obvious next thing is to give it llm audio responses in realtime

user->llm->audio_out+ditto_realtime

has anyone managed to do something like this?
can it run from a mac or a home computer? if it takes a big server, can it run many instances in parrallel and stream the video out?

thanks a lot and all the best
ps - found this out from https://github.com/Kedreamix/Awesome-Talking-Head-Synthesis