Personalized Head-Related Transfer Function Prediction Based on Spatial Grouping

TL;DR

This paper introduces a spatial grouping approach for predicting personalized head-related transfer functions (HRTFs) using neural networks, balancing accuracy and computational efficiency.

Contribution

The study proposes a novel spatial grouping method to improve HRTF prediction accuracy while reducing computational demands compared to angle-specific models.

Findings

The proposed method outperforms global models in HRTF prediction.

Spatial grouping improves prediction accuracy on both ipsilateral and contralateral sides.

The approach reduces computational resources needed for personalized HRTF modeling.

Abstract

The head-related transfer function (HRTF) characterizes the frequency response of the sound traveling path between a specific location and the ear. When it comes to estimating HRTFs by neural network models, angle-specific models greatly outperform global models but demand high computational resources. To balance the computational resource and performance, we propose a method by grouping HRTF data spatially to reduce variance within each subspace. HRTF predicting neural network is then trained for each subspace. Simulation results show the proposed method performs better than global models and angle-specific models by using different grouping strategies at the ipsilateral and contralateral sides.