Predicting Global Head-Related Transfer Functions From Scanned Head Geometry Using Deep Learning and Compact Representations

Authors: Yuxiang Wang, You Zhang, Zhiyao Duan, Mark Bocko

Required toolbox

SOFA toolbox\ SHtools\ PyColormap4Matlab\ Spherical Conformal Map\ spherical-cap-harmonics

Datasets used

HUTUBS HRTF database\ Included both the HRIRs and scanned head mesh of the subjects.

HRTF preprocessing

Process the HRTF into SHT representations in both frequency and temporal domains.

Head mesh processing

Isolate the ear region of each scanned head mesh, and process into SCH representations.

Deep learning model

Build the neural network to link the connections between the ear/torso geometry information to the HRTFs.