Comparison of the front-back confusion rate of individual and individualized head-related transfer functions
* Presenting author
The individualization of head-related transfer functions (HRTFs) is worthwhile if no individual data is available. One opportunity is to individualize an HRTF data set using anthropometric dimensions of the person to be adapted and principal components. However, the localization performance using an individualized data set could be worse than for an individual one. This is often based on the fact that a mismatch between the individual and individualized HRTFs is still remaining. In a listening experiment, this issue is investigated by the front-back confusion rate using individual and individualized HRTFs. The individualization approach for the HRTFs is based on an ITD model for low frequencies and the principal components in combination with the anthropometric dimensions. Hence, not only the individual and individualized HRTFs are tested but also the influence of the ITD model used and reconstructed HRTFs using principal components. Furthermore, to reduce the measurements for the anthropometric dimensions, the symmetry of HRTFs is investigated. The results for six tested directions show especially lower front-back confusion rate for frontal directions for individual HRTFs. For rear positions tested, all variants show similar rates.