Comment by bigzyg33k

advanced voice mode operates on audio tokens directly, it doesn't transcribe them into "text tokens" as an intermediate step like the original version of voice mode did.