Zubin Gahramani, VP of Google Research, talks about Google’s 1000 Languages initiative at the Google AI event in New York. Image: Sabrina Ortiz/ZDNET
At this week’s AI conference in New York, Google announced its intention to 10x its already vast language portfolio thanks to AI.
Google aims to create an artificial intelligence model that can support the world’s 1,000 most spoken languages to make information more accessible.
“Language is a fundamental part of communicating and understanding the world,” said Jeff Dean, senior researcher at Google. “But there are more than 7,000 languages spoken in the world, and only a few are well represented on the Internet today.”
Towards a universal voice model
Since the project is extremely ambitious, the project will most likely take many years to complete. However, Google is already working on achieving its goal.
According to a blog post, the tech giant has developed a Universal Speech Model (USM) that has been trained in over 400 languages, making it the most extensive speech model to date. Google is also partnering with communities around the world to provide voice data.
Google’s emphasis on expanding its language capabilities is not new. Google recently added 24 more languages to its Google Translate platform and enabled voice input for nine more African languages on Gboard.
Meta is also working on its translation AI model.
Google is also working with local governments, non-governmental organizations and academic institutions in South Asia to collect audio samples of the region’s various dialects.
Other big tech companies are also building big language models. In July, Meta announced an AI model called No Language Left Behind that can translate into 200 languages.
Meta has also made an effort to bring content to offline communities. The Meta AI model includes translations for 55 African languages, which is a significant improvement as less than 25 African languages are supported by widely used translation tools.