Tech giant Google has shared the progress made on the Universal Speech Model (USM) that it announced in November last year, as the AI war heats up.
Part of an ambitious plan to support the world's most spoken languages, USM is a state-of-the-art family of speech models with 2 billion parameters trained on 12 million hours of speech, 28 billion sentences of text spanning more than 300 languages.
While it hasn't reached the goal of 1,000 languages, it is an important foundation for Google.
Also Read | Civil rights audit at Google proposes better tackling of hate speech, misinformation
Google uses USM to generate live captions for videos on YouTube, and the search giant says that it can not only recognise "widely-spoken languages like English and Mandarin" but also "under-resourced languages like Amharic, Cebuano, Assamese, and Azerbaijani to name a few".
Also Read | YouTube promises to help creators make more money. Here are the details
The building blocks will serve as a foundation to build an even bigger language model.
"The development of USM is a critical effort towards realizing Google’s mission to organize the world’s information and make it universally accessible," wrote Google in the blog post.
Discover the latest Business News, Sensex, and Nifty updates. Obtain Personal Finance insights, tax queries, and expert opinions on Moneycontrol or download the Moneycontrol App to stay updated!
