Technology Desk: Artificial Intelligence i.e. AI is progressing rapidly. Now Meta has introduced a new speech-to-text model. Its name is Muse Voice Transcribe. This model can convert voice to text instantly. That is, your words will be visible written on the screen while speaking.
According to Meta, Muse Voice Transcribe supports more than 70 languages. It also includes the five major languages of India. These include Hindi, Tamil, Telugu, Malayalam and Kannada.
This model can also understand conversations in multiple languages simultaneously. For example, it will continue transcription even if the conversation switches from Hindi to English.
The specialty of this AI model is its real-time transcription capability. In this, you will not have to wait for the recording to finish. As you speak, the text will appear on the screen. Meta claims that it can recognize the voices of more than 20 people.
It can also transcribe recordings longer than an hour. This does not require any separate post-processing.
Muse Voice Transcribe is designed to be fast and accurate. It recognizes simple words immediately. It may take a little longer to understand difficult words. According to Meta, as of September 1, 2026, it was ranked first in a speech-to-text leaderboard.
Muse Voice Transcribe is now available through Meta's Model API. Currently it is being used for dictation in Mac's Meta AI and Muse Code. Meta has priced it at $3 per 1,000 audio minutes. This is equivalent to approximately $0.18 per hour.
This model can be very useful for multilingual countries. Especially users in India who speak Hindi and English together will benefit.
