A tool that detects when a person is speaking in audio.
VAD is like a motion light for sound. It wakes up when someone talks. Wind and silence get ignored.
It removes quiet parts before speech tools listen. This saves work and cuts false triggers.
ASR
VAD finds speech before ASR turns it into text.
Barge-in
VAD hears when a user starts speaking. It helps stop playback fast.
Full-Duplex Voice AI
VAD marks when speech starts and stops. This helps decide when to reply.