Contribution
Comparing Localization Accuracy with Generic, Estimated, and Individual HRTFs using Dynamic Binaural Synthesis
* Presenting author
Abstract:
In a study involving 24 participants, we examined the effectiveness of dynamic binaural synthesis using both individually measured and estimated head-related transfer functions (HRTFs) in comparison to generic HRTFs for vertical and horizontal localization accuracy. The study employed five different HRTF estimation techniques, high-frequency audiometry, proximal pointing, and individual headphone equalization. We used generic HRTFs from a KEMAR artificial head as a control, while individually measured HRTFs acted as a benchmark. Estimated HRTFs were derived from anthropometric data and 3D head models through methods such as adaptation of the interaural time difference, database selection, principal component-based synthesis, and numerical simulation. Results indicated that both measured and estimated HRTFs led to significantly smaller localization errors in elevation compared to KEMAR HRTFs, whereas horizontal errors did not show significant differences. Notably, the synthesis method emerged as the most effective, significantly reducing vertical localization errors for sound sources on and below the horizontal plane when compared to generic HRTFs.