Meta develops an AI model capable of transcribing and translating nearly 100 languages

Meta has developed an AI model, SeamlessM4T, capable of translating and transcribing nearly 100 languages across text and speech. This open-source model, in combination with a new translation dataset called SeamlessAlign, is considered a breakthrough in AI-powered speech translation. SeamlessM4T builds on previous models and frameworks, including the text-to-text machine translation model, No Language Left Behind, and the Universal Speech Translator. The new model was trained on data scraped from the web and claims to outperform current text and speech models. However, there are concerns regarding the potential biases present in the model and the loss of lexical richness with AI translations. Although Meta advises against using SeamlessM4T for certain applications due to potential misuse and mistranslations, the company aims to use this model to improve multilingual communication.