Google DeepMind previewed some advances of its video-to-audio research
Google DeepMind's Generative Media team has developed a video-to-audio (V2A) technology that can generate synchronized soundtracks for video input, with the ability to create multiple audio options guided by optional text prompts.