AI & Enterprise
Meta unveils real-time speech recognition model Muse Voice Transcribe
Meta unveiled Muse Voice Transcribe on Sept. 1, a real-time AI model that transcribes speech and distinguishes speakers. The company said it can handle conversations involving more than 20 people and recognise multilingual and mixed-language dialogue, including Korean. Developed by Meta Superintelligence Labs, it processes audio in 80-millisecond units and uses adaptive latency to balance accuracy and speed. Meta said it ranked first in an Artificial Analysis streaming test with a 3.1 percent word error rate.
