The number of virtual loudspeakers and the error for spherical microphone array recording and binaural rendering

摘要

Spherical microphone array recording and binaural rendering (SMARBR) is a spatial audio technique. It is implemented by firstly using Ambisonics (beamforming) scheme to decode the outputs of microphone array to the loudspeaker signals, and then further transferring the loudspeaker signals into binaural signals via virtual loudspeaker scheme. The present work evaluates the influence of number of loudspeakers on the error in binaural pressures for SMARBR. The results indicate that, for a given order of Ambisonics scheme, when the number of virtual loudspeakers exceeds the minimal requirement, on average, increasing number of virtual loudspeakers appropriately reduces the mean error around and above the high frequency limit imposed by spatial sampling theorem.