The technique that measures when audio in a spoken conversation should be classified as speech and when as silence. It forms the basis for turn detection en endpointing. Endpointing uses the VAD signal to determine that the other person has finished speaking.
See turn detection, endpointing
