Microsoft releases a three-model MAI speech stack
MAI-Transcribe-2-Streaming adds incremental real-time transcription, while MAI-Voice-2.1 and Flash add multilingual speech at published character rates. Microsoft documents the Azure models as public previews without an SLA.
By Benedict Felix · Updated 7 October 2026
What this changes for UK buyers
The component rates make a controlled pilot budget possible, but the temporary transcription price, preview status, reasoning layer, telephony and human escalation still determine production fit and complete cost.
What buyers should do
Run one consented end-to-end call flow with frozen audio and tools; measure partial and final errors, interruption recovery, accepted-call completion and every provider and review charge.
Primary announcement
Microsoft MAI speech-model launch ↗This is a dated source-linked buyer briefing. A vendor announcement does not prove UK account eligibility, an independent performance result or regulatory compliance. Open the primary source and confirm the exact plan before changing a production workflow.
Related profiles
Evidence standard · Commercial disclosure · Send a correction