A Binaural Synthesis with Multiple Sound Sources Based on Spatial Features of Head-related Transfer Functions

A low-complexity binaural synthesis technique is proposed for real-time immersive rendering of large number of sound sources. A multiple-input single-output system of the spatial features, extracted from head-related transfer function (HRTF) dataset, is constructed to render the sound sources. Sound sources are rendered using low-order approximation of this system. This technique is suitable for rendering large number of sound sources in the virtual auditory space with significantly lower computational complexity compared to the binaural synthesis techniques proposed earlier. The performance of the proposed technique is evaluated using perceptual objective measure along with subjective tests.