Updated

Scribe v3

ElevenLabs introduced Scribe v3, a speech-to-text model for recorded audio, with speaker diarization, multilingual support, and controls for terminology and number formatting. ElevenLabs reports fewer word errors than Scribe v2 across 102 languages in its evaluation.

By No Code MBA · Facts checked

Provider
ElevenLabs
Announcement date
Status at last check
Generally available

What changed

ElevenLabs describes Scribe v3 as its recorded-audio transcription model succeeding Scribe v2. It adds or improves speaker diarization, support for secondary languages, keyterm prompting for up to 1,000 terms, number formatting, and transcript editing. The company reports 27% fewer word errors than Scribe v2 across 102 languages; this is an ElevenLabs evaluation, not an independent result.

Access and availability

Available through the ElevenLabs Speech to Text API as model scribe_v3. ElevenLabs also documents use through its products and developer APIs.

Official access documentation

What you can build

Compare a representative set of recordings with your current transcription workflow, including multiple speakers and specialist terms, and review the transcript before using it downstream.

Limitations

The accuracy and diarization comparisons are ElevenLabs-reported evaluations and are not independent testing. The announcement does not specify regional access or API rate limits.

Sources and release history