Towards the design and evaluation of delay-based modeling of acoustic scenes in mobile augmented reality

ScatAR is an augmented reality (AR) audio application, where scattering delay networks (SDNs) efficiently generate and organize a reverberator based on room geometry scanned by an AR device. Here we present a perceptual evaluation of the application in terms of auditory object presence. Specifically, we investigate the personal preferences on the auditory registration of a sounding object (a drone transmitting speech) in a room in anechoic versus artificial reverberation conditions. The results indicate a marginal preference of the reverberation, but much less than what we have anticipated. We discuss possible reasons for this, and set to compare our SDN implementation with a Waveguide Web (WGW) simulation in AR. While the WGWs are computationally more demanding, they have a potential to provide more accurate second-order reflections, and therefore, better registering of audio-visual AR scenes.