Google’s Genie 4 lets you build worlds that sound and move like real life
It's the first Google world model that can generate audio.
Google has unveiled Genie 4, an AI world model capable of creating realistic simulations of physical environments with integrated sound generation. This marks Google's first real-time world model to offer native audio generation. Currently, the technology is in research preview and accessible only to a limited group of trusted testers.
While generative AI is often associated with video and image generation tools such as Gemini Omni or Nano Banana, world models are an area of intense research focus for many AI labs. These models construct an internal representation of a scene, employing physics, spatial relationships, and temporal progression to generate consistent outputs.
In contrast to language models that predict the next word based on token data, world models forecast how elements like space, time, gravity, and other physical forces will evolve within an environment.
Google's latest world model, Genie 4, builds upon its predecessor, Genie 3, introduced last year, and enhances it in several significant ways. By leveraging Gemini Omni, Genie 4 ensures that its simulations adhere closely to real-world appearances and behaviors. Additionally, Genie 4 boasts an impressive ability to maintain scene consistency for up to two minutes, a notable improvement from Genie 3's two-second limit.
One of the standout features of Genie 4 is its real-time adaptability, allowing users to modify the environment through natural language prompts. Users can adjust weather conditions, manipulate the camera, and introduce new people or objects into the scene. Despite these dynamic changes, Genie 4 ensures that the overall context remains coherent and consistent, providing users with an immersive and believable synthetic environment.
Written by urgent.news from Android Authority's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
