Facebook’s parent company, Meta platforms release artificial intelligence models that can translate speech into dozens of languages

On Tuesday, Meta Platforms, a Facebook subsidiary, unveiled an artificial intelligence model that can translate and transcribe voices in dozens of languages. This might serve as a foundation for technologies that facilitate real-time communication across language barriers.

In a blog post, the business claimed that by merging technologies from several models, their SeamlessM4T model could provide translations between text and speech in approximately 100 languages, and full speech-to-speech translation in 35 languages.

Mark Zuckerberg, the CEO, has stated that he foresees such technologies easing communication between people all over the world in the metaverse, a network of interconnected virtual worlds.

According to the blog post, Meta is releasing the model for public, non-commercial usage.

It’s no secret that Microsoft-backed OpenAI and Alphabet-owned Google have been facing stiff competition from the world’s largest social media firm this year, thanks to the rush of free AI models they’ve published, including a massive language model dubbed Llama.

According to Zuckerberg, Meta benefits from an open AI ecosystem since it can more successfully crowd-source the development of consumer-facing tools for its social platforms this way than by charging for access to the models.