Meta is boosting its speech and translation AI with a fresh program
Meta has launched a new Language Technology Partner Program, seeking volunteers to provide transcribed speech, translations, and written text. Contributors will collaborate with Meta's AI experts and receive training on open-source language tools. The initiative aims to collect data on underrepresented languages, with a particular focus on preserving endangered Inuit languages in Canada.
Joining Forces for Language Inclusivity
Meta has partnered with UNESCO to advance open-source AI by gathering speech and text data, expanding the understanding and translation of marginalized languages worldwide. The Language Technology Partner Program invites individuals to contribute at least 10 hours of transcribed speech recordings, pre-translated sentences, and written content in their native languages.
This data will play a crucial role in Meta's AI development, enhancing its language translation capabilities to include often overlooked languages. Participants will have the opportunity to collaborate directly with Meta's AI experts, assisting in integrating new languages into open-source speech and translation models. Additionally, contributors will receive practical training through workshops on Meta's open-source language tools.
Canadian Territory Leads the Way
Nunavut, a Canadian territory known for its endangered Inuit languages, has already pledged its support for the initiative. The Language Technology Partner Program is open for applications until March 7, 2025, welcoming volunteers to join the effort.
Advancing Machine Translation
As part of the program, Meta has introduced an open-source machine translation benchmark to evaluate the performance of language translation models. Developed by linguists, the benchmark currently supports seven languages and is accessible on the Hugging Face platform for community contributions.
Meta's language program builds upon the success of previous projects, such as "No Language Left Behind" and "Massively Multilingual Speech," which have demonstrated the company's commitment to language diversity. The expansion of Meta's language capabilities includes features like auto-translation for content creators, with plans for innovative tools like voice translation for Instagram Reels.
Industry Trends and Commitment
Google and other tech giants are also prioritizing the preservation of endangered languages, as seen with Google's inclusion of Inuktut in its translation services. Meta presents its language initiatives as altruistic endeavors, emphasizing the importance of linguistic diversity while leveraging advancements in speech and translation AI for business growth.
