A New Model for Room Acoustic Prediction and Auralization

The development of two room acoustic prediction models and auralization methods are described. The first prediction model is a combination of two well known algorithms with some extensions; the image source model and ray-tracing. The second model consists of an entirely new algorithm based on a unified treatment of diffuse and specular reflection, and addresses many of the drawbacks of previously reported models. The new model applies approximate cone-tracing where diffuse reflection is handled by split-up of cones striking diffusing surfaces. The algorithm utilizes properties of diffuse reflection to avoid the very long calculation times associated with a brute force implementation of cone split-up. To further decrease the computational power required, reflection growth extrapolation is applied to enable to the use of a low number of secondary cones. The algorithm adapts dynamically to the acoustical properties of a given environment and assigns most processing power to where it is needed, i.e. to the early part of the echogram and to strong reflections. The properties of the new algorithm makes it especially well suited to create audible simulations - auralization - since it creates millions of weak diffuse reflections and thereby a smooth reverberation. To facilitate binaural post-processing of the high number of reflections generated, reflection decimation, based on the limitations of human hearing is applied.