Meta AI · Shipped
Muse Voice Transcribe release date
Released
Sep 6, 2026
What we know
Real-time speech transcription model processing 80-millisecond chunks with speaker diarization and sentence boundary detection. Positioned as foundation for personal AI agents on wearable devices.
Every claim
2 from 2 sources- The DecoderFrom the labSep 6
Meta's Superintelligence Labs released Muse Voice Transcribe, a real-time audio transcription model processing speech in 80ms chunks.
- TechmemeFrom the labSep 1
Meta launches Muse Voice Transcribe, a real-time audio perception model for streaming automatic speech recognition.