Scalable immersive audio for virtual environments

Abstract

This thesis discusses the history of game sound and how it developed to become increasingly more immersive and realistic by accounting for the spatial aspects of sound. It then compares several 3D sound technologies before continuing on to present a custom spatial sound system that uses the highly parallel nature of modern graphics processing units (GPUs) to process large amounts of data for simulated sound propagation. In a real-time virtual reality game environment, the system is able to handle dynamic geometry and movable sound sources in full 3D at interactive rates. To test the implementation, I developed a VR-compatible video game for desktop and mobile platforms with an emphasis on play-by-sound gameplay mechanics. Results demonstrated low overhead on all devices tested when used with the appropriate pathfinding backend (CPU/GPU). Additional testing yielded that the GPU part of the system is capable of processing several thousand 3D paths at interactive rates.