A Ray Tracing Simulation of Sound Diffraction Based on the Analytic Secondary Source Model
Masashi Okada, Takao Onoye, Wataru Kobayashi · IEEE Transactions on Audio Speech and Language Processing · 2012
This paper describes a novel ray tracing method for solving sound diffraction problems. This method is a Monte Carlo solution to the multiple integration in the analytic secondary source model of edge diffraction; it uses ray tracing to calculate sample values of the integrand. The similarity between our method and general ray tracing makes it possible to utilize the various approaches developed for ray tracing. Our implementation employs the OptiX ray tracing engine, which exhibits good acceleration performance on a graphics processor. Two importance sampling methods are derived from different aspects, and they provide an efficient and accurate way to solve the numerically challenging integration. The accuracy of our method was demonstrated by comparing its estimates with the ones calculated by reference software. An analysis of signal-to-noise ratios using an auditory filter bank was performed objectively and subjectively in order to evaluate the error characteristics and perceptual quality. The applicability of our method was evaluated with a prototype system of interactive ray tracing.