Scribe v2 Realtime
Scribe v2 Realtime is the live-transcription branch of Scribe. It supports language changes during a session and can use previous transcript text when a connection restarts.
Partial and committed text
A partial transcript can change as more speech arrives. A committed segment contains the fuller context. Applications can decide when to commit a segment, or use voice activity detection to commit after a pause.ElevenLabs: how streaming Scribe transcripts work
Practical guidance. Display partial text for immediate feedback, but use committed text for saved notes and downstream actions. Otherwise an early recognition guess can become a permanent record.ElevenLabs: how streaming Scribe transcripts work
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | API, Provider product |
| Model identifiers | scribe_v2_realtime |
| Latency measure | About 150 ms median to a committed transcript; network and application time excludedElevenLabs latency measurement method |
Related models
- Scribe v2
ElevenLabs · Scribe
Batch speech-recognition model for long recordings, structured transcripts and multilingual audio.
Launch date API · Provider product - Cue
ElevenLabs · Eleven alignment
Character-level forced-alignment model for placing supplied text on an audio timeline.
Launch date not confirmedAPI