Archive · 2 Sep 2026
Meta Launches Muse Voice Transcribe with Indian Language Support
Meta, a big tech company, has made a new tool called Muse Voice Transcribe.
This tool can write down what people say as they speak, in real-time.
It works with many languages, including five big Indian languages: Hindi, Tamil, Telugu, Malayalam, and Kannada.
The tool can also tell who is speaking among many people and handle long recordings.
It's really good at understanding when people switch between languages.
Meta says this tool is the best at what it does.
It's available for people to use and costs a little money to use a lot.
This tool is helpful for things like writing down speeches, helping people who can't hear well, and making voice assistants better.
Meta Superintelligence Labs launched Muse Voice Transcribe, a real-time audio perception model supporting five major Indian languages.
The model offers streaming transcription, speaker separation for over 20 voices, and native code-switching without post-processing.
Muse Voice Transcribe is trained on over 70 languages, with 25 validated at launch, and ranks first on the Artificial Analysis streaming speech-to-text leaderboard as of September 1, 2026.
It features adaptive delay, dynamically adjusting latency for each word based on difficulty, and is available via Meta's Model API priced at $3 per 1,000 audio minutes.
The model is already in use for dictation in Meta AI for Mac and Muse Code, aiming to improve real-time transcription in multilingual markets like India.
- Who
- Meta Superintelligence Labs
- What
- Launched Muse Voice Transcribe, a real-time audio perception model
- When
- September 1, 2026
- Where
- Global release, with specific focus on Indian languages
- Why
- To improve real-time transcription, speaker separation, and multilingual support, particularly in multilingual markets like India
- Model Name
- Muse Voice Transcribe
- Developer
- Meta Superintelligence Labs
- Supported Indian Languages
- Hindi, Tamil, Telugu, Malayalam, Kannada
- Total Languages Trained
- 70+
- Validated Languages at Launch
- 25
- Speaker Separation Capacity
- 20+ voices
- API Pricing
- $3 per 1,000 audio minutes
- Leaderboard Ranking
- 1st on Artificial Analysis streaming speech-to-text leaderboard (as of September 1, 2026)
