Wave-Based Geometrical Acoustics for Accurate and Efficient Sound Rendering

Abstract

Sound rendering has many important applications in both engineering software tools and virtual reality systems. However, the complexity of the calculations has led to a gap between research for accurate acoustic modeling techniques and techniques used for efficient sound rendering in performance critical and engineering applications such as virtual reality systems and acoustical simulation tools, resulting to respective trade-offs in both research fields. In this thesis, we attempt to bridge the gap by presenting a calculation model that can deliver accurate calculation results for a wide range of cases in computation times suitable for practical use. Our model combines geometrical acoustics, a technique which is widely used for efficient sound rendering applications with wave-based equations which allow for the accurate calculation of wave-based phenomena like resonances, noise cancelling and room modes. Our work is incorporated in a commercially available software application as a demonstration of the practical applicability of the calculation model. The model incorporates wave-based calculations for sound reflections, sound diffractions, atmospheric absorption, and atmospheric refraction. Algorithms for detecting the relevant sound paths are introduced. We validate the performance of our model by running simulations on several different setups and comparing the results with sound measurements and theoretical predictions. Our results show an improvement in accuracy when compared to other widely used engineering models. Also, our model proves capable of accounting for wave-based phenomena like sound diffractions, room modes and the seat-dip effect. Furthermore we improve the efficiency of geometrical acoustics tracing algorithms by looking into how the tree traversal type affects the performance of such algorithms, and by proposing methods that use an intelligent prioritization of the tree traversal to improve the computation time of sound path detection. We show that the type of the tree traversal can affect the performance of algorithms based on the image source method, with differences being of perceptual significance. We introduce a novel algorithm using a prioritized ray tracing technique and we show improvements when compared to a non-prioritized version of it as well as other popular algorithms. At last, we use prioritization techniques to demonstrate improvements in calculating room acoustics parameters in engineering applications.