Influence of parametrically-decoded First-order ambisonics reproduction in cinematic virtual-reality-based soundscape evaluation

Abstract

Increasing consumer availability of first-order ambisonics (FOA) microphones have enabled virtual evaluations of soundscapes with head-tracked binaural playback through headphones. Known issues with FOA binaural decoding techniques, such as timbral and spatial artefacts, however, have marred its advantages. Parametric decoding methods, such as directional audio coding (dirAC), high angular resolution plane-wave expansion (HARPEX), and the recently proposed coding and multidirectional parameterization of ambisonic sound scenes (COMPASS) have noted significant improvements in sound quality over time-invariant linear processing of FOA. Whereas subjective assessments of the parametric methods have been traditionally focused on musical media, few studies have focused on the evaluation of outdoor acoustic scenes. Hence, the differences between parametrically and linearly decoded FOA to head-tracked binaural is explored here through subjective assessments of outdoor acoustic scenes. Participants evaluated the perceived spatial sound quality and the soundscape quality of four reproduced outdoor scenes in cinematic virtual reality through a head-mounted device. The spatial quality was evaluated based on bipolar attributes related to the overall impression and impression of the acoustic environment during head movement. The results show that the evaluation of soundscape quality was independent of the reproduction methods. However, a notable improvement in the perceived spatial quality was observed in the parametric reproductions in terms of directionality and distinctiveness of sound sources.