Microsoft's new transcriber starts writing before you finish speaking.
Microsoft released MAI-Transcribe-2-Streaming on 1 October, its first live speech-to-text model. It returns draft words about 100 milliseconds after hearing audio and finishes 0.13 seconds after you stop. Artificial Analysis ranks it first for streaming accuracy at a 2.5% error rate, and it covers 60 languages.
Why it mattersLive captions, dictation and voice agents feel broken when the text lags behind the talker. An introductory $0.54 an hour until year end puts real-time transcription inside small projects.
Read in the app — free, in 9 languages
Related stories
ElevenLabs is giving students a free year of its reading voice.2026-10-05An American lab is about to give away a China-class open model.
2026-10-05A video site now gives away the best-scoring open translation model.
2026-10-05Half the people on a short video call thought the AI was human.
2026-10-04Nvidia posted a new multimodal model and told nobody about it.
2026-10-04