If there are periods of silence in an active audio stream, the turn will end after speech.endDetected fires. We cannot initiate a turn start preemptively - that's for the service to decide for us. Therefore, on each turn.end, a new request id should be generated and made available for new audio packets to include when sending. This new request id will then initiate the next turn start for us.
If there are periods of silence in an active audio stream, the turn will end after
speech.endDetected
fires. We cannot initiate a turn start preemptively - that's for the service to decide for us. Therefore, on eachturn.end
, a new request id should be generated and made available for new audio packets to include when sending. This new request id will then initiate the next turn start for us.