Deepmind showcases the capabilities of its video-to-audio (V2A) artificial intelligence, which can add sound to videos based on commands. Providing various usage directions, such as adding sound to AI-generated videos or enhancing soundless old videos.
V2A supports user prompts, allowing for versatile sound tone adjustments, even within the same video. For instance, adding music to a specific scene or simply background noise.
While AI-generated sound from text or images has been presented before, V2A’s advantage lies in its understanding of videos to create synchronized sound fitting seamlessly. Key features include generating dialogue synced with characters’ facial expressions in the video.
Deepmind continues to limit the use of these AI capabilities within the creative community initially, similar to other models unveiled at Google I/O. They emphasize the importance of ensuring security before opening up to a wider audience.
TLDR: Deepmind demonstrates the advanced video-to-audio AI technology, allowing for customizable sound additions to videos with synchronized dialogue and facial expressions. They prioritize security before broader release.
Leave a Comment