LiveKit Community

does STT models like Deepgram once return transcribed text if VAD detects. If VAD doesn't detect any voice, will Deepgram wont return any text?

Thanks for reaching out, Deepak. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.


Thanks — good question. You're asking whether Deepgram's STT returns transcripts only when VAD detects voice, and whether it returns nothing during silence.

Short answer: Yes. Deepgram processes audio continuously but only emits transcripts when speech is detected. If VAD (or Deepgram's internal speech-detection) doesn't detect voice, it will not return transcription events during those silent periods. You can enable options like `vad_events` (to get explicit start/stop events) and configure `no_speech_timeout`/endpointing to control behavior around silence and stream termination.
 Sources: <https://docs.livekit.io/recipes/transcriber|Transcriber | LiveKit docs> | <https://blog.livekit.io/open-source-realtime-multimodal-ai/#/portal|An open source stack for real-time multimodal AI> | <https://blog.livekit.io/meet-kitt/#/portal/signup|Live Conversations with AI using ChatGPT and WebRTC>