Individual Differences in Training Naive Listeners to Localize Spatial Audio in Virtual Reality

It has been widely believed that a key factor in creating realistic spatial audio in virtual reality (VR) is the head-related transfer function (HRTF), which is unique to each individual, but costly to measure for widespread use. This study investigates the effects of HRTF personalization and training on sound localization accuracy in VR. Two experiments were conducted: Experiment 1 compared naive listeners and those who underwent brief training on localization tasks using personalized versus generic HRTFs; Experiment 2 used a within-subject design to assess training effects over two sessions. Results show that accurately localizing sound can be a difficult task for many participants the first time. Training significantly improves localization accuracy, reducing errors and confusions, and enabling many initially non-sensitive listeners to perceive spatial audio effectively. Although HRTF personalization yielded a statistically significant benefit, the effect was small, primarily improving elevation perception at extreme angles. These findings suggest that generic HRTFs combined with user training may suffice for most VR applications.