MAI-Transcribe-1

Microsoft's speech recognition model: 25 languages, the lowest word error rate among competitors at launch and 2.5x faster than Azure's previous offering. It holds accuracy through background noise and overlapping speech. Priced from $0.36 per hour of audio.

Анонс 02.04.2026. Публичное превью в Microsoft Foundry и MAI Playground; поэтапно включается в голосовой режим Copilot и Microsoft Teams.

Understands

audio

Produces

text

Contents verified with the vendor: 2026-08-18

Included in subscriptions

No catalogue entry marks this model as part of a plan yet. It will appear here once the plan contents are confirmed with the vendor.