Universal Speech Model
A general-purpose automatic speech recognition model released by Google, covering speech recognition capabilities for more than 300 languages. It is designed to advance the broad accessibility of AI language technology.
Timeline (last 90 days)
Google's Universal Speech Model (USM) already covers speech recognition for more than 300 languages.
Traditional STT relies on Hidden Markov Models and Gaussian Mixture Models, while recent end-to-end deep learning models have significantly improved recognition accuracy and multilingual support capabilities.
In recent years, end-to-end deep learning models such as OpenAI's Whisper and Google's Universal Speech Model have significantly improved recognition accuracy and multilingual support capabilities.
All Facts (4)
Google's Universal Speech Model (USM) supports more than 100 languages.
70%UnverifiedGoogle's Universal Speech Model (USM) already covers speech recognition for more than 300 languages.
50%UnverifiedTraditional STT relies on Hidden Markov Models and Gaussian Mixture Models, while recent end-to-end deep learning models have significantly improved recognition accuracy and multilingual support capabilities.
50%UnverifiedIn recent years, end-to-end deep learning models such as OpenAI's Whisper and Google's Universal Speech Model have significantly improved recognition accuracy and multilingual support capabilities.
50%