The audio source the bot publishes its voice through. VERIFY: captureFrame(frame) is async.
captureFrame(frame)
Optional
Duration (in ms or s) of queued audio currently buffered in the source, if supported.
Pushes one PCM frame onto the published track.
Drops all audio still queued in the source (used to flush on barge-in / interruption).
The audio source the bot publishes its voice through. VERIFY:
captureFrame(frame)is async.