Washington, US – Meta has announced the launch of a groundbreaking new multilingual artificial intelligence system, capable of supporting over 1,600 languages with a framework built for potential expansion to cover more than 5,400 global languages.

The new system, which reportedly utilises learning within the context without previous training is being hailed as a strategic turning point for the company in the field of multilingual AI.

Crucially, the technology dramatically lowers the barrier to linguistic inclusion. It allows developers to integrate entirely new languages using only a few audio and text examples, eliminating the need for full model retraining. Meta asserts that this unique capability makes it the most scalable speech-known system to date.

The company has underscored its commitment to reducing barriers to innovation and supporting global linguistic diversity by making all resources for the multilingual AI system freely available. The package released includes a set of language models, a multilingual voice representation model, and a substantial database encompassing more than 350 rare languages.

This transparent approach, driven by the reach of Meta’s widespread social platforms, enables the model to be used without restriction for both research and commercial projects. Developers are granted direct access to download the forms and data from platforms such as GitHub and Hugging Face, complete with easy programming interfaces for system integration into commercial applications.