Supports real-time voice translation for nearly 100 languages with a delay of around 2 seconds, capable of replicating complex features such as speech pauses, tone, and speed. Utilizes a non-autoregressive architecture to support long sequence translations. Open-sourced four translation models along with relevant data and tools, adding audio watermarks to prevent model misuse and reduce translation toxicity.

1. Meta's newly open-sourced translation model SeamlessM4T supports translation in 100 languages and voices. 2. SeamlessM4T integrates the advantages of multiple translation models, including NLLB and MMS. 3. SeamlessM4T is trained on 270,000 hours of aligned speech and text data.

{1:Meta has released the world's largest open-source multimodal translation model, supporting 100 languages and capable of recognizing regional dialects. 2:SeamlessM4T is the first integrated AI translation model that supports multimodal translation tasks including speech-to-text, speech-to-speech, text-to-speech, and text-to-text. 3:SeamlessM4T integrates previously released translation models by Meta, such as NLLB and MMS, using 270,000 hours of aligned speech and text data.}