Meta launches Muse Voice Transcribe with 3.1% word error rate for real-time dictation

Published Sep 1, 06:30 PM UTC

Editorial responsibility

Prepared by: WorldAttention Editorial Systems
Human review: No named human review is recorded for this page.
Automated steps: Source reporting is collected, normalized, translated or condensed automatically when needed.
Page status: Automatically published source-backed update

How automation and AI are used · Report a correction or rights issue

Meta has released Muse Voice Transcribe, a new real-time voice dictation model that achieves a 3.1% final-transcription word error rate with adaptive delay, significantly outperforming competing models. Unlike conventional speech-to-text APIs, Muse processes audio in 80-millisecond chunks and dynamically decides when to emit text, when to wait, and when a speaker change or turn ends, functioning as a real-time perception layer for voice agents. The model is available through the Meta Model API, Meta AI for Mac, and Muse Code. This release represents a major advancement in streaming voice transcription technology, offering lower latency and higher accuracy for developers building voice-enabled applications.

Source report

Meta has released Muse Voice Transcribe, a new real-time voice dictation model that achieves a final-transcription word error rate of just 3.1% with adaptive delay — significantly outperforming competing models.

Key Innovation

Muse goes beyond conventional speech-to-text systems. Rather than simply transcribing speech, it learns to:

Determine when to wait before committing a word
Recognize when a speaker changes
Identify when a turn is actually over

All of these capabilities are integrated within a single streaming model, making Muse function more like a real-time perception layer for voice agents than a traditional speech-to-text API.

How It Works

Muse processes audio in 80-millisecond chunks
After each chunk, it decides whether to emit text or continue listening

Availability

Muse Voice Transcribe is available through:

Meta Model API
Meta AI for Mac
Muse Code

Source

rohanpaul_aiNeutral / independent

Original headline: Love this, another huge release from Meta. Lunched Muse Voice Transcribe for real-time voice dictation, with the lowest 3.1% final-transcription word error rate with adaptive delay. which is significantly ahead of competing models. - The big

Original language: EN

Part of this Story

Meta releases Muse Voice Transcribe with 3.1% WER, undercutting rivals at $3 per 1,000 minutes

See the full Story: all sources and camp coverage