Effect of head-tracking artefacts on externalization and localization in non-individualized binaural synthesis

When listening to binaural sound sources over headphones with non-individualized HRTFs, it is common to perceive the sound inside the head rather than externalized. Several studies have shown that large head movements coupled with head-tracking can improve the perception of externalization. In that context, head-tracking provides several additional perceptual advantages. Head-tracking allows to adapt the simulated direction of a virtual sound based on the head movements of the listener, so as to simulate a stable sound source in space. By allowing the differential integration of interaural time and level differences, head-tracking also reduces the occurrence of front-back errors associated with the cone of confusion and thus improves localization performance. This study focuses on the case of a non-individualized anechoic spatialization to which are superimposed early reflections (up to 10-ms) with the help of various simulated head-tracking algorithms. The study includes two subjective listening experiments. The first one aims at evaluating the effect of head-tracking artefacts on the potential improvement in externalization brought by dynamic binaural synthesis. The simulated artefacts consist of a large latency or an amplitude mismatch. Different levels of direct-to-reverberant ratios (DRR) were tested. The results suggest that even with the tested head-tracking artefacts, the perception of externalization is improved compared to the case without head-tracking and comparable to the reference head-tracking. A second experiment aimed at evaluating to which extent those artefacts can affect localization performance. The results suggest that even with a large latency, listeners remain as accurate as with the reference tracking. However their performance is degraded in the case of head-tracking with amplitude mismatches