Multi-Resolution Sound Rendering

Abstract

Point-based multi-resolution representations have been used successfully for rendering highly complex three dimensional scenes in real-time. In this paper, we apply this paradigm to sound rendering: A hierarchy of stochastic sample sound sources is used to approximate complex sound environments (containing a large number of sound sources, such as a football stadium), allowing for interactive real-time walkthroughs. Additionally, the proposed technique can be used for observer-dependent auralizations of simple approximations of global sound propagation. Typical applications of the technique are in virtual reality and computer games, especially to complement established output-sensitive algorithms for rendering visual content.