ElevenLabs Scribe v3 turns recorded or live speech into text for meeting notes, interview transcripts, subtitles, and voice applications.
| Model | When to use it | Where to use it |
|---|---|---|
| Scribe v3 | Recordings such as meetings, podcasts, interviews, and videos. It returns word-level timestamps and can distinguish up to 32 speakers. | ElevenLabs website or Speech to Text API |
| Scribe v3 Live | Live captions, meeting notes, and voice support. It streams text as people speak and can label speakers. | Realtime WebSocket API or ElevenAgents |
The standard web transcription tool uses Scribe v3 by default. After uploading audio or video, click a word in the transcript to jump to that point in the recording, correct the text, rename speakers, and export it. In the API, you can tell Scribe to write numbers as digits or words and provide additional language hints for mixed-language audio. See the model documentation and web transcription guide.
A caller might stop halfway through a phone number to check the remaining digits. A fixed silence threshold can finalize the segment during that pause. Scribe v3 Live's turn prediction uses the audio and the conversation context to estimate whether the speaker has finished. See the realtime transcription guide.
commit_strategy=turn_prediction.The free web plan includes 12 minutes per month of recorded transcription. ElevenLabs lists $0.20 per audio hour for Scribe v3 and $0.22 per audio hour for Scribe v3 Live on its API pricing page. Use the rates for your chosen entry point when estimating costs. See the web plans and API prices.
ElevenLabs Scribe v3
Ready to try it? Open the official website.