Abstract
This publication includes techniques to obtain universal head related transfer functions (HRTFs) by adapting generic HRTFs. The adaptation is performed by iteratively perturbing and fine tuning the latent space representation of the generic HRTFs. The tuning objective is to minimize the binaural localization error produced by the adapted HRTFs across a population of users. To estimate the binaural localization error, the system is configured to train a person-agnostic neural network binaural localization model that can estimate the direction of a sound source from binaural audio.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Hersek, Sinan and Shin, Dongeek, "System For Generating A Universal Transfer Function For Rendering Spatial Audio Content", Technical Disclosure Commons, (October 03, 2025)
https://www.tdcommons.org/dpubs_series/8671