Abstract
This disclosure describes a system for generating head-related transfer functions (HRTFs) in real time for 3D video conferencing systems. HRTFs, which are unique to each user, often should be modeled with accuracy to provide an immersive user experience when rendering spatial audio in a 3D video conference. However, systems face challenges generating accurate HRTFs in real time. This system generates real-time HRTFs using existing 3D video conferencing infrastructure to generate a 3D model of a user’s head and torso using multiple cameras. From the 3D model, an accurate HRTF can be generated and provided to a spatial audio rendering module. The system can repeat this process in real-time with imperceptible latency to maintain the immersive user experience. Machine learning models can be used to generate the 3D model and HRTF.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Guo, Jian and Shin, Dongeek, "Real-time Head-related Transfer Function Generation for 3D Video Conferencing", Technical Disclosure Commons, (October 17, 2024)
https://www.tdcommons.org/dpubs_series/7447