Meta is expanding its AI translation efforts by partnering with UNESCO to enhance speech recognition and translation for underrepresented languages . Many Indigenous languages lack digital support, making them harder to preserve and integrate into modern communication tools. With this initiative, Meta aims to bridge that gap by improving AI-powered translation systems.

As part of its Language Technology Partner Program , Meta invites researchers, governments, and linguistic experts to contribute transcribed speech recordings (at least 10 hours), large written texts, and bilingual sentence pairs . By gathering this data, the company hopes to train AI models that accurately recognize and translate languages that mainstream technology has overlooked.

Supporting Indigenous Language Preservation

Meta has already partnered with the government of Nunavut , a remote Canadian territory where people speak Inuktitut and Inuinnaqtun . This collaboration will help develop AI-driven translation tools tailored for these Indigenous languages, making them more accessible in digital spaces.

UNESCO has long advocated for Indigenous language preservation. This initiative aligns with its mission under the International Decade of Indigenous Languages , which focuses on revitalizing and protecting native languages worldwide.

Setting a New Standard for AI Translation

Meta is not only collecting language data but also launching BOUQuET , an open-source benchmarking tool designed to evaluate AI translation models . Unlike existing tools, BOUQuET will feature expertly crafted sentences to ensure AI-generated translations capture accuracy and cultural context .

By releasing BOUQuET as an open-source project, Meta hopes to encourage researchers and developers to collaborate on building more inclusive and advanced translation models .

AI-Powered Translation on the Rise

Meta continues to invest in AI-driven language tools. Last year, it introduced a Reels dubbing tool that automatically translates videos and syncs lip movements to different languages. The tool was initially launched in English and Spanish , but Meta plans to expand its reach globally.

Meta also rolled out its AI Assistant in 43 countries , supporting more than a dozen languages. This assistant integrates AI-powered translation, making cross-language communication smoother and more accessible .

By combining AI innovation with linguistic expertise , Meta and UNESCO are taking a major step toward breaking language barriers and ensuring underrepresented languages thrive in the digital world.