Events

Date: 11 March 2026   Time: 15:00 - 16:00     

Location: Engineering G2

Abstract

XR applications such as gaming require high-quality 6DoF acoustic rendering to achieve perceptual immersion. In Virtual Reality (VR), room geometry and materials are known, whereas in Augmented Reality (AR) they must be inferred from measurements. In both settings, the renderer must accurately reproduce early reflections and late reverberation while updating in real time as the listener moves in 6-DoF. Geometric acoustics methods like ray tracing and recent Neural Acoustical Field approaches can reconstruct room impulse responses throughout a space. However, they typically rely on costly time-varying convolution for rendering. In this talk, I present delay-network-based artificial reverberators as a computationally efficient alternative. I show how Feedback Delay Networks can be tuned directly from geometric parameters in VR or from measured impulse responses in AR. The resulting approach enables real-time 6DoF updates without storing large databases of impulse responses, making physically informed acoustic rendering practical for interactive XR applications.

Bio

Orchisama Das is a Postdoctoral Research Associate at King's College London. In 2022-2024, she was Senior Audio Research Scientist at Sonos Inc., London. In 2021-2022 she was a Research Fellow at the Institute of Sound Recording, University of Surrey, Guildford, UK. She received the B.Eng. degree in Instrumentation and Electronics Engineering from Jadavpur University, India in 2016, and a PhD degree from the Center for Computer Research in Music and Acoustics at Stanford University, USA in 2021, during which she interned at Tesla and Meta Reality Labs. Her research interests are immersive audio rendering, room acoustics modeling and real-time artificial reverberation.