HRTF Personalization Based on Pinna Shape Estimation by Standardized Scanning with Handy 3D Scanner

As one of the three-dimensional (3D) sound field reproduction systems, binaural system that controls 3D sound image localization using only headphones with head-related transfer functions (HRTFs) has been studied. In order to get accurate reproduction, it is necessary to use the listener's own HRTF or a suitable HRTF. For this reason, we have proposed a conventional method that selects the suitable HRTF from an HRTF database for the listener using spectral notch frequencies which were estimated from five pinna images taken by a stereo camera. However, there are some problems such as the low stability and long time spent caused by the pinna shape estimation from every image which needs to be taken in actual angle respectively. In order to solve these problems, we proposed a new method that estimates spectral notch frequencies from a 3D pinna model acquired by standardized scanning with a handy 3D scanner. Through two objective experiments evaluating estimation accuracy of the individualized HRTFs and time spent on measurement, the results revealed that the proposed method provided a simpler and accurate HRTF personalization compared with our conventional method. Otherwise, we found out the scanning standards including scanning way and scanning time with handy 3D scanner.