Meta releases Muse Voice Transcribe: Real-time ASR with 20+ speaker diarization
shuchaobi · x · 2026-09-02
Meta has launched Muse Voice Transcribe, described as the first real-time audio perception model from its Superintelligence Labs. Key features include:
- Real-time streaming ASR: Ultra-low latency transcription.
- Speaker Diarization: Real-time identification of 20+ speakers in a single session.
- Multilingual Support: Native handling of 25+ languages with seamless code-switching.
- Context Biasing: Improved accuracy via language, keyword, and context biasing.
Meta claims top rankings on Artificial Analysis for streaming speech-to-text and public diarization benchmarks. A live web demo is available.
More from Multimodal
- One prompt turns Claude Code into a full motion-graphics video pipeline with Remotion — aziz4ai · 2026-10-18
- Flux 3 Text-to-Image Can't Make Realistic Faces at All, Reddit User Reports — EstablishmentFun6629 · 2026-10-18
- MiniMax H3 Ref-to-Video: 15s Shot from 3 Images + Vocals, 1088p Latent Refinement — michael_goldenhour · 2026-10-18
- Dev builds fully local AI tool generating HD 3D textures with Qwen, may sell for $5 — TinfoilTricorn · 2026-10-18
- 14-Agent AI Film Pipeline on One RTX 5090: Fixing Cut Defects Between Clips — developervkmp · 2026-10-18
- 4MP Text-to-Image Comparison: Qwen Image 2.1 Turbo vs MOSS vs Krea 2 vs Flux 3 Klein — FORNAX_460 · 2026-10-18