ElevenLabs just released Scribe v3 and Scribe v3 Live (scribe_v3 and scribe_v3_live), and it would be awesome to see them added as cloud voice model options alongside Scribe v2 and Scribe v2 Realtime.
Why Scribe v3 would be a great addition:
Significant accuracy boost: Scribe v3 delivers 27% fewer errors than Scribe v2 (across 102 languages) with improved number formatting, multilingual/code-switching support, and lower diarization error.
Much better real-time dictation: Scribe v3 Live delivers 43% fewer errors than Scribe v2 Realtime, with improved turn detection and keyterm library support.
Same API pricing for Scribe v3: On ElevenLabs’ API pricing page (https://elevenlabs.io/pricing/api), Scribe v3 is priced at $0.20/hour—the exact same rate as Scribe v2 (and Scribe v3 Live is $0.22/hour).
Straightforward migration: Both models use the standard ElevenLabs Speech-to-Text API (scribe_v3 and scribe_v3_live), and would also be great to support for users bringing their own ElevenLabs API key.
Are there any plans to add Scribe v3 and Scribe v3 Live in an upcoming release?
Thank you for building such a great tool!