Final Thesis

Perception-based real-time scheduling strategies for room acoustic simulations

Master Thesis of Palenda, Pascal

During the past decades, fast room acoustic simulation tools have been established. These tools are based on the geometric acoustic approach considering sound as rays instead of waves. Their simulation results, called room impulse response (RIR) or filter, allow to auralize a room in a virtual environment. In other words, a user can get the impression of being in this room while listening to a virtual scene e.g. using headphones. In order to make a virtual scene more realistic, a real-time interaction such as walking through the room is desired. Nevertheless, this requires an update of the filter in a reasonable amount of time. For a maximum efficiency, it is desired to only do a filter update if audible changes are to be expected. For this purpose, an intelligent, acoustically motivated scheduling system is developed in this thesis. This scheduler should be based on thresholds that correspond to an audible change of the RIR e.g. if the user moved a certain amount of distance. In the course of this work, potential variables for filter variables are detected and corresponding thresholds are investigated. Then the scheduler framework is integrated into the current system.