Yes. You can access live transcripts during an active conversation in a few supported ways:
If you are the client connected to the agent:
Use the Agent WebSocket (wss://api.elevenlabs.io/v1/convai/conversation?agent_id=...) and listen for server-to-client events. You’ll receive user_transcript (finalized ASR per user utterance), agent_response, and agent_response_correction in real time. (elevenlabs.io)
If you need to observe an ongoing conversation from outside (supervisor/QA use case):
Use the Real-time monitoring WebSocket wss://api.elevenlabs.io/v1/convai/conversations/{conversation_id}/monitor to stream live conversation events, including transcripts, responses, and corrections. Note: enterprise-only; requires Agents Write scope and must be enabled in the agent’s settings; streams text/metadata (no raw audio). (elevenlabs.io)
If you need partial (interim) transcript updates while capturing audio yourself:
Use Scribe v2 Realtime Speech-to-Text over WebSocket. It streams partial_transcript events as you speak and committed_transcript(_with_timestamps) when segments finalize. (elevenlabs.io)
For after-the-call data, you can also receive the complete transcript via post-call webhooks. (elevenlabs.io)
how should i let the agent listen quitely and only response at certain point, is this can be done by just prompts, but i need to enable certain settings or tools
agile hazel
fair oak