On February 7, 2025, Meta announced its new Language Technology Partner Program, which looks for collaborators to help improve and expand its open-source language technologies, particularly AI translation tools.
The Fundamental AI Research (FAIR) team’s latest initiative at Meta (side note: FAIR previously stood for Facebook AI Research) prioritizes what it calls “underserved languages” and aligns with the International Decade of Indigenous Languages program at the United Nations Educational, Scientific and Cultural Organization (UNESCO).
Specifically, the company is looking for partners that can contribute speech recordings with transcriptions, written text, and translated sentences for low-resource languages. The Government of Nunavut, Canada, for example, has already joined the program, pledging to share data in the Inuit languages Inuktitut and Inuinnaqtun.
Those interested in contributing to the project should be able to provide over 10 hours of speech recordings with transcriptions and more than 200 sentences of written text, along with translated sentences.
The work consists of integrating low-resource language datasets into speech recognition and machine translation (MT) models. In return, collaborators will have the opportunity to work directly with Meta’s research teams, access technical workshops, and learn how to leverage the resulting open-source models.
2024 Slator Pro Guide: Translation AI
The 2024 Slator Pro Guide presents 20 new and impactful ways that LLMs can be used to enhance translation workflows.
As part of the No Language Left Behind (NLLB) project, launched in 2022, Meta began collaborating with UNESCO and Hugging Face on an open-source MT engine. The resulting translation interface was announced during the United Nations General Assembly week in September 2024.
Meta also made the Meta Massively Multilingual Speech (MMS) interface available in Hugging Face to support the same UNESCO program. For this project, Meta’s own massively multilingual/multimodal translation models support transcription for over 1,000 languages and are capable of zero-shot speech recognition.
In the same partner program announcement, Meta unveiled the BOUQuET open-source machine translation benchmark. This is a standard seven-language evaluation tool intended to help evaluate AI translation for massively multilingual text-to-text machine translation systems.
A few weeks before this latest announcement, the FAIR team published in January 2025 an update in the Nature journal about the SeamlessM4T system, a set of models capable of automatic speech recognition (ASR), text-to-text translation (T2TT), speech-to-text translation (S2TT), text-to-speech translation (T2ST), and speech-to-speech translation (S2ST).
Slator Translation as a Feature (TaaF) Report
The Slator Translation as a Feature (TaaF) Report is a vital and concise guide on how AI translation is becoming an integral feature in enterprise technology.
The FAIR team states in Nature that while the No Language Left Behind project broadened T2T translation coverage to more than 200 languages, speech translation should be prioritized because “unified S2ST models are far from achieving similar scope or performance. This disparity could be attributed to many causes, but audio data scarcity and modelling constraints remain key obstacles.”
Meta also stated that with these projects, “ultimately, our goal is to create intelligent systems that can understand and respond to complex human needs, regardless of language or cultural background.”
