The New Stack says Meta’s Muse Voice Transcribe tracks 20-plus speakers and uses reinforcement learning to learn how long to listen before each word.
The New Stack reports that Meta’s Muse Voice Transcribe leads a streaming speech-to-text benchmark, beating OpenAI and Google at real-time transcription. According to the outlet, it can track 20-plus speakers, while reinforcement learning teaches it how long to listen before each word.
864
1 post, first seen 7d ago
The New Stack says Meta’s Muse Voice Transcribe tracks 20-plus speakers and uses reinforcement learning to learn how long to listen before each word.
The New Stack reports that Meta’s Muse Voice Transcribe leads a streaming speech-to-text benchmark, beating OpenAI and Google at real-time transcription. According to the outlet, it can track 20-plus speakers, while reinforcement learning teaches it how long to listen before each word.
The New Stack@thenewstackMeta just beat OpenAI and Google at real-time transcription
The New Stack@thenewstackMeta just beat OpenAI and Google at real-time transcription
