Sabrina Ortiz/ZDNET
AI-generated video has been advancing rapidly, with leading tech developers racing to build and commercialize their own models. We’re now seeing the rise of tools that can generate strikingly photorealistic video from a single prompt in natural language. For the most part, however, AI-generated video has had a glaring shortcoming: it’s silent.
No longer. At its annual I/O developer conference on Tuesday, Google announced the release of Veo 3, the latest iteration of its video-generating AI model, which also comes with the ability to generate synchronized audio.
Also: Everything announced at Google I/O 2025: Gemini, Search, Android XR, and more
Imagine you prompt the system to generate a video set inside a busy subway car, for example. Veo 3 can produce the video, along with AI-generated ambient background noise to add to the sense of realism. You can even prompt it to generate audio of human voices, according to Google.
The model also reportedly specializes in simulating real-world physics and lip-syncing, making it a potentially valuable tool for filmmakers and advancing Google’s broader mission of bringing usable AI to creative industries. It’s available now for Gemini Ultra subscribers in the US. It can also be accessed through Flow, Google’s new AI-powered filmmaking tool, which was also unveiled at I/O this week.
A major technical challenge
Veo 3 represents one of the first models from a major tech developer that can synchronize AI-generated video and audio. Meta’s Movie Gen, released in October, is another. Some other tools, like Runway’s Gen-3 Alpha, come with features that enable AI-generated audio to video in a post-production process, but the concurrent generation of the two requires the compute and resources of a major force like Google.
Also: 8 best AI features and tools revealed at Google I/O 2025
Building AI models capable of generating synchronized video and audio has been a thorny technical challenge and an active area of research across the AI industry. Both AI-generated video and AI-generated audio are distinct technical challenges, and fusing them introduces a whole new dimension of complexity. Here’s a demo of Veo 3.
For one thing, video is a series of still frames, whereas audio is a continuous wave. Syncing the two therefore requires models that can operate across these two modalities, accounting for the vastly different timescales in which they operate.
Also: Google Flow is a new AI video generator meant for filmmakers – how to try it today
An AI model fusing video with sound must also be able to dynamically account for variables like material, distance, and speed. A car driving at 100 miles per hour sounds a lot different than one traveling at 10 miles per hour; a horse walking on cobblestones sounds different than one that’s walking on grass.
Get the morning’s top stories in your inbox each day with our Tech Today newsletter.
