Transcribe audio
ElevenLabs Scribe
Speech-to-text API that turns audio and video files or live streams into transcripts with timestamps and speaker labels.
At a glance
- Provider
- ElevenLabs
- Category
- Transcribe audio
- Type
- Hosted API
- Pricing model
- Pay per hour of audio, $0.22 batch and $0.39 realtime1
- Free tier
- Free plan with 10k credits a month13
- Authentication
- API key in the
xi-api-keyheader; single-use tokens for client-side use6 - Connect via
- REST API, WebSocket (realtime), SDKs, CLI, local MCP server2
- Read from the provider's pages
- 2026-11-04
Pricing
- Pricing model: Pay per hour of audio sent; monthly subscription plans include hours
- Free tier: Free plan with 10k credits a month
- Card required for free tier: Not published
- Official MCP server: Yes, local
- Address or package:
uvx elevenlabs-mcp(PyPI packageelevenlabs-mcp); no hosted address for transcription
- Who creates the account: A person signs in on the dashboard to create a key; the docs also describe a CLI flow that opens a browser to sign in or create an account
- Where to get a key: Dashboard at
https://elevenlabs.io/app/api/api-keys - Other requirements: Not published
- REST API:
https://api.elevenlabs.io/v1/speech-to-text - MCP server: Local only:
uvx elevenlabs-mcp - OpenAPI spec:
https://api.elevenlabs.io/openapi.json
- Max duration: 10 hours per file (standard mode)
- Max duration (other page): 1 hour (multichannel mode)
- Minimum audio length: 100 ms
- Data location: US by default; EU, India and Singapore on Enterprise
- Retention: Inputs and outputs of STT requests are logged by default; ZRM is Enterprise only
- Used for training: The privacy policy says personal information is used to train models; opt-out in account settings
Read from ElevenLabs’s own pages on 2026-11-04. Every number links to the page a value came from. For agents, the same facts as markdown: /tools/transcribe-audio/elevenlabs-scribe.md