2026.10.27 Realtime SaaS

New

Oak 1, a medical model, is now available for real-time transcription. It delivers improved medical keyword recognition compared to melia-1, with the same base accuracy. To get started, set model to oak-1 when opening the WebSocket connection.

Fixes

Melia 1 and Oak 1 models:

• Transcripts in languages written without spaces — such as Chinese and Japanese — no longer place a space between every word. Words now follow their own language's delimiter, with a space only where the language changes.

• Fixed word timings drifting after a forced end of utterance.

• Fixed a case where a valid WAV file could be misidentified as another format during audio preprocessing. Affected files now transcribe correctly.