Synthesis of Near-Field HRTFs by Directional Equalization of Far-Field Datasets
* Presenting author
HRTFs are virtually independent of sound source distance in the far field, but vary significantly in the near field. The change of an HRTF when a sound source shifts in distance can be described by a distance variation function (DVF). To synthesize near-field HRTFs, it is a common method to apply DVFs to far-field HRTFs. In this study, we present a modified version of our recently proposed method for spatial upsampling of sparse HRTF datasets. The approach is based on a spectral equalization of the sparse HRTF dataset with a directional rigid sphere transfer function (STF), spatial upsampling of this dataset by an inverse spherical harmonics transform on a dense grid, and a spectral de-equalization of the processed dataset with the same STF. In the modified implementation, near-field HRTFs can be synthesized by applying a STF for a point source in the near field as the de-equalization function instead of a STF for a plane wave in the far field. The method is pretty similar to the DVF method, but has the advantage that parallax effects can be appropriately considered. To evaluate our approach, we compare synthesized near-field HRTFs to measured ones, focusing on binaural cues and spectral characteristics.