Creation of representative head-related impulse responses for binaural rendering of moving audio objects

Hoshito K, Nishiguchi M, Watanabe K, Abe K, Ishikawa T, Usami H. Creation of representative head-related impulse responses for binaural rendering of moving audio objects. In: Express Paper. Audio Engineering Society; 2024. Paper 255. Available from: https://aes.org/publications/elibrary-page/?id=22714

@inproceedings{Hoshito2024_22714,
  author = {Hoshito, Kazuki and Nishiguchi, Masayuki and Watanabe, Kanji and Abe, Koji and Ishikawa, Tomokazu and Usami, Hikaru},
  title = {{Creation of representative head-related impulse responses for binaural rendering of moving audio objects}},
  note = {Paper 255},
  year = {2024},
  month = sep,
  publisher = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=22714}
}

TY  - CPAPER
TI  - Creation of representative head-related impulse responses for binaural rendering of moving audio objects
AU  - Hoshito, Kazuki
AU  - Nishiguchi, Masayuki
AU  - Watanabe, Kanji
AU  - Abe, Koji
AU  - Ishikawa, Tomokazu
AU  - Usami, Hikaru
M1  - Paper 255
PY  - 2024
DA  - 2024/09/27
UR  - https://aes.org/publications/elibrary-page/?id=22714
PB  - Audio Engineering Society
LA  - en
AB  - The considerable computational complexity involved in convoluting head-related impulse responses (HRIRs) must be addressed to achieve a highly realistic 3D audio reproduction in virtual reality (VR) or augmented reality (AR) through binaural rendering. An algorithm has been proposed to reduce this complexity by distributing audio signals into predefined representative directions through panning. Only these distributed signals are convoluted with the corresponding HRIRs. In this study, we explore a method to generate representative HRIRs through training by utilizing a full-sphere HRIR set. This approach simultaneously considers smooth transitions and minimal degradation during rendering for both moving and static audio objects. Compared with conventional panning, the proposed method reduces the average distortion by approximately 47%, while slightly increasing the rendering runtime complexity.
ER  -