**[Audio Model Company Enters the World Model Space, Noiz Completes Tens of Millions in Seed Round Financing]** According to monitoring by Beating, audio AI company Noiz has completed a seed round financing of tens of millions of yuan, with the latest investor being GSR Ventures. Unlike most audio models, it does not plan to continue focusing on music generation, speech recognition, or TTS. Instead, it aims to enable AI to understand the physical world behind sounds. For example, upon hearing a door close, the model would not only recognize "the door closed" but also determine the door's material, the force used, the distance of the sound, and how the room's structure altered the sound. Noiz is specifically collecting data that includes spatial, material, distance, and directional information. Ultimately, it aims to create something more akin to a "sound-based world model": given the current environment and ongoing events, the model would directly simulate what should be heard at that moment. This way, audio models could evolve from being dubbing and sound effect tools to being used in real-time gaming, virtual worlds, and robotics, enabling machines to understand what is happening around them through sound. [Original Link]