Google marks major milestone towards its 1,000
Google has announced a major milestone in its development of an AI model that can operate in 1,000 languages. The company’s new Universal Speech Model (USM) is a family of AI models that have been trained on 12 million hours of speech and 28 billion sentences of text. This is the first time that such a large-scale dataset has been used to train a speech recognition model.
The USM is based on the open-source BERT models developed by Google. BERT is a powerful tool for natural language processing (NLP) that can be used to develop a wide range of language applications. The USM will enable developers to build speech recognition applications that can operate in multiple languages.
Google has been working on developing a multi-lingual AI model for some time. In 2018, the company released the Cloud Speech-to-Text API, which supported over 100 languages. The new USM will significantly expand the number of languages that can be supported by Google’s speech recognition technology.
The release of the USM marks a major milestone in Google’s efforts to make its AI technology accessible to more people around the world. The company is committed to making its products and services available in as many languages as possible. With the release of the USM, Google is one step closer to achieving its goal of supporting 1,000 languages.
