Sù is a musical interface for direct manipulation of a real-time generative audio system built around RAVE. The project explores how a technically complex latent synthesis model can become a tool for creative control and audio spatialisation.
The work began as a greenfield research project during my research internship at Dolby’s Advanced Technology Group in Australia in 2025.
I was drawn to the appealing concept of “feedback” in creative control design, inspired by Tom Mudd’s research. The possibility of nonlinear interaction bringing unpredictability, immediacy, and entangled agency to our experiences with technological interfaces seemed exciting, opening up space for the future design of AI-powered creative tools.
With supervision and mentoring from David Gunawan, Benjamin Southwell, and David Cooper, we explored the implementation of feedback loops within a latent synthesis model. This approach combines direct control with a degree of randomness or chance, inviting users and creators to explore and discover sounds during content creation.
The core sound engine, including the RAVE component, was implemented in Max/MSP, alongside a mixed reality interface for object-based audio spatialisation developed in Unity using C# and the Meta OpenXR SDK, running on a Meta Quest 3 headset.
