AI-generated Answer for rianna_03418
Short answer: Yes to both—(1) you can retrieve/download past conversations, and (2) you can train your own model externally and plug it into ElevenLabs.
How to download previous conversation data
- List and page through prior conversations: GET /v1/convai/conversations. (elevenlabs.io)
- Fetch a specific conversation (includes full transcript, metadata, and flags indicating available audio): GET /v1/convai/conversations/:conversation_id. (elevenlabs.io)
- Download the call audio recording: GET /v1/convai/conversations/:conversation_id/audio. (elevenlabs.io)
- Receive transcripts/analysis (and optionally base64 audio) automatically after each call via Post‑call Webhooks. (elevenlabs.io)
- If you used the Speech‑to‑Text service directly, you can also retrieve those transcripts by ID: GET /v1/speech-to-text/transcripts/:transcription_id. (elevenlabs.io)
Training and using your own model
- ElevenLabs Agents support Custom LLMs, so you can train/host your own model (or bring your own OpenAI key) and connect it via an OpenAI‑compatible endpoint. (elevenlabs.io)
- If you’re building your own text agent, you can add ElevenLabs’ voice UX (turn‑taking, TTS/STT) on top of it using the Speech Engine guide. (elevenlabs.io)
- The platform also documents patterns for integrating more complex external agents behind the Custom LLM interface. (elevenlabs.io)
Notes
- If Zero Retention Mode is enabled for compliance, conversation data may not be stored for later retrieval; review the HIPAA/Zero Retention guidance. (elevenlabs.io)
- You can manage whether ElevenLabs uses your submissions to improve its models in Data use settings (this doesn’t restrict your ability to export your own data). (help.elevenlabs.io)
sand rivet
cobalt dome