Technology
Microsoft ships a streaming transcription model and two multilingual voices
Microsoft AI released MAI-Transcribe-2-Streaming, which returns provisional transcripts while audio is still arriving and commits a stable version once a speaker pauses. The company says it covers 60 languages with automatic detection, produces its first hypotheses in the low hundreds of milliseconds, and costs $0.54 per audio hour on an introductory basis through the end of the year. It also launched MAI-Voice-2.1 and a faster Flash variant, both supporting 23 languages with one cross-language voice, priced at $22 and $15 per million characters.