Exploring AI Audio Models in Soundwalking with Broader Audiences

The development of generative AI offers new possibilities for music composition and production. However, it could undermine the creative process and obscure the generation steps between input and output, and lead to homogenous results. To enable broader audiences to engage with generative AI models, we adopt soundwalking - an embodied listening and compositional practice - as a design principle, and apply meaningful human control to its AI integration. Together, these frameworks can encourage the player's exploration and understanding of the sonic outcome. This demo presents an immersive soundwalking experience created using an AI audio synthesis model. The audience walks in the virtual environment as their movements transform into soundscapes in real time. The interactive system is built in an accessible and playful way for a broader gallery audience to engage with AI models through embodied listening.