Navigation Performance Effects of Render Method and Latency in Mobile Audio Augmented Reality

This paper describes a pilot study and main experiment that assess user performance at navigating to spatialised sound sources using a mobile audio augmented reality system. Experiments use a novel outdoor paradigm with an application-relevant navigation task to compare perception of two binaural rendering methods under several head-turn latencies. Binaural rendering methods examined were virtual, 6-speaker, first-order Ambisonic and virtual 12-speaker VBAP techniques. This study extends existing indoors research on the effects of head-turn latency for seated listeners. The pilot study examined the effect of capture radius (of 1, 2, 3, 4 and 5 metres) on mean distance efficiency for a single user’s navigation path to sound sources. A significant performance degradation was found to occur for radii of 2 m. The main experiment examined the effect of render method and total system latency to head-turns (176 ms minimum plus 0, 100, 200, 400 and 800 ms) on mean distance efficiency and subjective stability rating (on a scale of 1-5), for 8 participants. Render method significantly affected distance efficiency and 800 ms of added head-turn latency significantly affected subjective stability. Also, the study revealed a significant interaction effect of render method and head-turn latency: Ambisonic rendering didn’t significantly affect subjective stability due to added head-turn latency, while VBAP rendering did. Thus, it appears rendering method can mitigate or potentiate stability effects of head-turn latency. The study also exemplifies that the novel experimental paradigm is capable of revealing statistically significant performance differences between mobile audio AR implementations.