Home / Technology / Meta's AI Listens: Real-Time Audio Model Masters 20+ Speakers
Meta's AI Listens: Real-Time Audio Model Masters 20+ Speakers
2 Sep
Summary
- New AI model handles dictation for over 20 speakers simultaneously.
- It seamlessly transcribes multiple languages and code-switching.
- Model trained on 70+ languages, validated on 25 at launch.
Meta has unveiled Muse Voice Transcribe, a groundbreaking real-time audio model. This new AI technology excels at dictation and transcription, supporting over 20 speakers and multiple languages simultaneously. Its adaptive delay feature allows it to predict tokens, enhancing accuracy.
The model demonstrates remarkable capabilities, including handling mid-sentence code-switching and long sessions. Trained across more than 70 languages, with 25 validated at launch, it promises robust multilingual support. This release positions Meta as a key player in advanced audio AI.
Muse Voice Transcribe is the latest innovation from Meta Superintelligence Lab. While Google recently launched a similar model, Meta's offering is available to developers via Muse Code and its Model API for $3 per 1,000 audio minutes. A demo is also available on Meta's research blog.
