Neowin®

Meta has announced its new generative AI model, Voicebox, which can perform tasks such as audio editing, sampling and styling, that it wasn't specifically trained to do through learning.

Meta has announced its new open-source AI model called ImageBind. It's a multimodal system that can interoperate across six different data modes such as text, image, video, 3D, thermal, and motion.

Meta has released a new AI model that identifies objects within images just by selecting them, or by writing a prompt word which will make the tool select each object that matches the prompt.

Meta has built an language model that translates across 200 languages better than the competition at an average of 44%. It has up to 54.5 billion parameters and has been open-sourced on GitHub.