Meta AI Updates: September 2, 2026

1. Meta Releases Muse Voice Transcribe for Real-Time Speech

Meta. Muse Voice Transcribe is Meta Superintelligence Labs’ first real-time audio perception model, running streaming ASR, speaker diarization for 20 or more speakers, and endpointing in a single pass with no required post-processing. It handles more than 25 languages natively with code-switching, and supports language, keyword, and context biasing to steer recognition. Meta reports first place on Artificial Analysis for streaming speech-to-text and on public diarization benchmarks as of September 1, 2026, consolidating three systems that production pipelines normally stitch together. Source