Spatial Audio for WebXR: Perceptual Evaluation of Sound Localization Technologies on the Browser

This study presents a comparative analysis of three open-source spatial audio systems for WebXR: Web Audio API’s PannerNode with HRTF, PannerNode with equalpower, and Google Resonance Audio for A-Frame. We developed a WebXR application featuring a spherical array of $\mathbf{4 2}$ virtual speakers to evaluate sound localization accuracy across different directions and positions. In our within-subjects experiment, twelve participants identified the perceived source location of pink noise stimuli by selecting speakers within the virtual environment. Results showed that PannerNode with HRTF significantly outperformed both alternatives in correct target identification, while Google Resonance Audio produced significantly larger error distances. Performance varied by direction, with PannerNode-HRTF particularly effective for back-positioned sources, and position-specific advantages observed for different systems. Despite objective performance differences, participants reported no significant differences in cognitive workload or simulator sickness between implementations. These findings provide empirical evidence that spatial audio implementation significantly affects localization accuracy in WebXR without impacting user comfort. Our methodology establishes a framework for evaluating spatial audio in web-based immersive environments while addressing browser compatibility constraints. This research offers practical guidance for WebXR developers selecting appropriate audio technologies based on their application requirements.