MicrosoftMAI Transcribe
MAI-Transcribe-2-Streaming
MAI-Transcribe-2-Streaming is a public-preview API that transcribes audio as it arrives.
Preview
What the release provides
It returns intermediate and final transcript segments. Integrations can use a Realtime-compatible WebSocket or Azure Speech SDK. The model supports automatic language detection and is served from listed Azure regions.MAI-Transcribe-2-Streaming API overview
How to use it well
Practical guidance. Keep partial text separate from final text in the interface and stored transcript. Review how corrections propagate to captions. If the application starts an action from a partial phrase, it must handle the speaker changing or correcting that phrase.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Preview |
| Access | API |
| Model identifiers | MAI-Transcribe-2-Streaming |