© Shutterstock
Google’s AI lab came through with groundbreaking research that could radically transform media creation. Their latest invention, dubbed “video-to-audio” or V2A, has the potential to make full movies, TV shows, and other visual content with perfectly synchronized sound, all generated by artificial intelligence.
DeepMind revealed they have developed technology that can pair AI-generated music, sound effects, and even dialogue with AI-produced video. This marks a major step forward, as current video-creating AI models can only output silent clips. DeepMind realized bringing these generations to life requires soundtracks, and their solution understands raw pixel data to autonomously select audio befitting each scene.
While still working out kinks like dialogue generation and handling video artifacts, DeepMind acknowledged this technology could become the missing link in unlocking fully AI-crafted movies. The implications are huge – goodbye to traditional production and licensing?
Their cautious approach makes sense given ongoing AI safety debates. With photorealistic video generation now public, deepfakes loom as a menace; pairing sound could escalate that threat. DeepMind aims to address such risks through techniques like their video watermarking SynthID tool. They’re also getting creator feedback to ensure positive impacts on media jobs at stake in AI’s rise.
