Towards Transfer-Plausibility for Evaluating Mixed Reality Audio in Complex Scenes

Abstract

The evaluation of mixed reality audio is typically approached under the paradigms of either authenticity or plausibility. While the first refers to the identity of a real and a virtualized sound source, the latter measures the degree of belief in cases where no direct reference is available. We refer to transfer-plausibility as the ability of a virtualized source to stand alongside multiple real sound sources. We present a perceptual experiment where listeners detect and identify a sound source as being virtualized using dynamic non-individualized binaural rendering under varying scene complexity. Scene complexity is controlled by a varying number of loudspeakers. We demonstrate that the presented methodology mitigates ceiling effects, typically encountered in authenticity and plausibility tests.

Publication
AES International Conference on Audio for Virtual and Augmented Reality (AVAR)

Listening Test