RENDERING BINAURAL SIGNALS FOR MOVING SOURCES

Authors
L M Gomes, W D'Andrea Fonseca, D Carvalho, P H Mareze

Immersive audio systems can provide humans the auditive perception of being in a different environment and/or surrounded by virtually any type of sound source. The principle for achieving such immersion scenarios is to simulate the relevant acoustic information that may persuade a listener’s brain. In order to do so, one creates virtual and similar sound waves which collide with a binaural receiver. To create a sound scene, direct sound, reflections, and interactions with the ground and objects are important. However, considering the human psychoacoustic model, some reduction of the problem is possible. Furthermore, the relative movement of the source and receiver must be taken into account. Considering a moving sound source, the perceived frequency content of a listener will be affected by its relative speed. Depending on how fast the changes occur, the Doppler Effect can be relevant and useful information in order to provide the sensation of movement. In the present study, it is proposed a method to render binaural signals that create scenes where the static listener identifies a sound source passing-by (like an ambulance, for example). The following sections outline the methods and theories about the sound propagation model (used in the simulations) and HRFTs (Head-Related Transfer Function) adjustments to render the binaural audios. The present work relies on digital signal processing techniques. The time-windowing algorithm (based on the source-receiver angle) and the HRTF processing techniques, such as extrapolation, VBAP (Vector Base Amplitude Panning), and Bilinear interpolation are also presented and accordingly discussed. Finally, subjective testing reveals interesting data concerning the perception of the distinct binaural audio signals tested.