Meta uses Bible translations and audio recordings to improve the speech model and now supports more than a thousand languages

Meta has greatly advanced speech recognition technology, and released a single multilingual speech recognition model MMS (Massively Multilingual Speech) model, which can recognize more than 4,000 spoken languages, and this number is 40 times that of currently known technologies. MMS has also enabled text-to-speech, and speech-to-text technology, to expand from about 100 languages ​​to 1,107.