Meta builds a language translator for the world to socialize in the metaverse

Meta CEO Mark Zuckerberg gave an overview of ongoing projects at Meta and what the Metaverse will look like in the coming future during the livestream of the event “Inside the Lab: Building for the Metaverse with AI “.

Zuckerberg revealed that Meta is working on building an AI-powered “universal speech translator” for the Metaverse that will work for everyone in the digital world. This project aims to streamline the interactions of users who speak different languages ​​in the digital world using AI.

“The goal here is instant speech-to-speech translation in all languages, even the most widely spoken ones; the ability to communicate with anyone in any language,” Zuckerberg said during his presentation. “It’s a superpower that people have always dreamed of and AI is going to deliver it in our lifetime,” said he added, as reported by engadget.com.

According to Meta’s blog, 20% of the world’s population does not speak the languages ​​covered by current translation tools. Other underrepresented languages ​​are hard for AI tools to learn because they don’t have enough scripts written for machine learning (ML) algorithms.

For languages ​​like English, Mandarin, or Spanish, web apps and tools already provide high-quality translation technology, but for other languages, it doesn’t, the blog says.

Meta’s universal language translation plan will be executed in two parts. In the first part, Meta develops a “No Language Left Behind” program, a translation system capable of learning all spoken languages ​​even if there is not much text available to learn.

“We are creating a single model that can translate hundreds of languages ​​with industry-leading results and most language pairs, from Austrian to Ugandan to Urdu,” Zuckerberg said during the live broadcast. direct.

In the second, Meta aims to create an AI Babelfish, a universal speech translator that aims to build systems that directly translate speech in real time from one language to another without the need for an intermediate written component.

However, the scarcity of data is Meta’s biggest hurdle when it comes to translation tools. Current text translation systems rely on learning from millions of annotated data sentences. Therefore, only a few languages ​​that dominate the web are available for translation. Expanding to other languages ​​creates a need to find ways to acquire and use training examples from languages ​​with little presence on the web, Meta wrote in his blog.

With these projects, Meta aims to improve translations so that people who speak different or even lesser-known languages ​​can socialize in Metaverse.

(Edited by : Thomas Abraham)