4 ms·

I really like the realtime aspect, with interrupts and silence detection happening as well as in open ai’s advanced voice mode