International Conference on Artificial Reality and Telexistence Eurographics Symposium on Virtual Environments (2018) G. Bruder, S. Cobb, and S. Yoshimoto (Editors) Geometrical Algorithms for Real Time Sound Rendering Using Intelligent Prioritization Panagiotis Charalampous 1 and Despina Michael-Grigoriou 1,2 1 GET Lab, Department of Multimedia and Graphics Arts, Cyprus University of Technology 2 Research centre on Interactive media, Smart systems and Emerging technologies(RISE), Cyprus Abstract Geometrical algorithms have been the main subject of research in the field of real time sound rendering. These algorithms are variants of the image source and ray tracing algorithms, enhanced with improvements that speed up substantially their performance. The fundamental concepts behind the improvements achieved up to now was the reduction of the processed information and the acceleration of the actual processing. In this paper, we show how altering the traversal method affects significantly the algorithm’s performance. These optimizations alter its behavior, providing better results for real time purposes. We separate the techniques into three major categories and we propose a stochastic Monte Carlo algorithm which involves optimizations based on prioritization. 1. Introduction Sound is an important component of an immersive 3D virtual envi- ronment as serves significant functions in such environments such as providing the sense of location and orientation, creating emo- tions and enhancing user immersion. Hence, delivering realistic sound in 3D virtual environments is a non-negligible task for such applications. As a result, a significant amount of research is de- voted in delivering high quality audio in virtual reality and games applications. One of the areas of research in this field is the real- istic simulation of sound propagation in 3D space for interactive applications. In this paper, we present a method for improving the perfor- mance of geometrical acoustics algorithms used in sound render- ing based on the concept of prioritization. Essentially, we propose improvements to a widely used tracing algorithm which alter its behavior and improve its performance for real time purposes. Our improvements are based on the concept of decomposing the tracing problem into a tree traversal problem and prioritizing the traversal of the tree in a way that more valid specular reflections are detected over the same amount of time. Our prioritization technique is based on automatically adjusting tracing termination criteria during run- time instead of explicitly setting them beforehand. We validate the algorithm based on paths detected, relative sound pressure and re- verberation time calculated over a fixed amount of time. We com- pare the algorithm with other widely used techniques. Experimen- tal executions of the algorithm indicate that this technique provides perceptually important improvements on almost all cases. 1.1. Related Work We classify the related work based on the way they handle process- ing to improve performance. The categories we propose are pro- cessing reduction algorithms, processing acceleration algorithms and processing prioritization algorithms. Tree pruning algorithms are those algorithmic techniques that result to the reduction of the size of the tree, therefore improving execution times substantially. Most of the advances in sound prop- agation calculations fall under this category. Most notable tech- niques are visibility tracing [Mec02], beam tracing [FTC * 04], frus- tum tracing [CLT * 08] and ray tracing [KSS68]. All these tech- niques share the same concept, that of tracing geometrical primi- tives through the 3D environment and detecting which geometrical objects are visible from other objects. By neglecting non visible objects from further processing, the tracing results to a pruned tree, which includes only nodes which are geometrically visible from their parent nodes. Beam and frustum tracing techniques are based on the im- age source method introduced by Alen and Berkley [AB79] for rectangular rooms and extended by Borish for arbitrary polyhe- dral [Bor84]. Since then, several improvements were suggested, like Vörlander’s hybrid ray tracing/image source implementation, Mechel’s validity criteria [Mec02] and Schröder’s binary space par- titioning [SL06]. Beam tracing is a method that uses beams to accurately prune the tree of possible image sources, by only considering the visi- ble leafs of each parent node. Beam tracing is currently considered as the fastest commonly used geometric room acoustics modeling technique [LSLS09]. Recent developments in this area include the c  2018 The Author(s) Eurographics Proceedings c  2018 The Eurographics Association.