The head-related transfer function (HRTF) describes how sound propagates from a source to both ears. It contains important information for human sound localization and is essential for achieving binaural reproduction, a spatial audio technology for headphones. However, HRTFs vary significantly among listeners, and using another person’s HRTF for binaural signals often results in localization errors. Therefore, to achieve high-quality binaural reproduction, it is crucial to obtain an individual’s HRTF using a simple setup. We are investigating techniques such as upsampling HRTFs from a small set of measurements and estimating an individual’s HRTF from ear shape parameters using machine learning.
@article{Ito:IEEE_OJSP2025,author={Ito, Yuki and Nakamura, Tomohiko and Koyama, Shoichi and Sakamoto, Shuichi and Saruwatari, Hiroshi},title={Spatial Upsampling of Head-Related Transfer Function Using Neural Network Conditioned on Source Position and Frequency},journal={{IEEE} Open Journal of Signal Processing},year={2025},volume={6},pages={1109-1123},doi={10.1109/OJSP.2025.3613132},}
@inproceedings{Niu:WASPAA2025,author={Niu, Ryan and Koyama, Shoichi and Nakamura, Tomohiko},title={Head-Related Transfer Function Individualization Using Anthropometric Features and Spatially Independent Latent Representation},booktitle={Proceedings of {IEEE} Workshop on Applications of Signal Processing to Audio and Acoustics ({WASPAA})},month=oct,year={2025},doi={10.1109/WASPAA66052.2025.11230978},}
