EFFECT OF SPATIAL SAMPLING APPROACHES ON VIRTUAL HIGH ORDER AMBISONICS

Ambisonics is a sound spatialisation technology that aims at physically reconstructing the sound field. When the ambisonics signals are not reproduced by loudspeakers, but instead synthesized to binaural signals suitable for headphone reproduction, the concept of virtual ambisonics or binaural synthesis is introduced. Virtual high order ambisonics (HOA) may be applied since there is no difficulty in arranging a large number of loudspeakers. In order to use head-related transfer functions (HRTFs) in binaural synthesis, the sampling of virtual loudspeaker of HOA should be set up to match the sampling distribution of HRTFs. However, the existing sampling approaches of measured HRTFs may be inefficient for HOA. In this work, based on the calculated HRTFs of a scanned KEMAR artificial head, the effects of sampling approaches of HRTFs on binaural synthesis are analysed. Results show that the nearly uniform sampling is more efficient in virtual HOA and binaural synthesis, but the HRTFs should be obtained first via spatial interpolation or numerical simulation. 1