Mixing time prediction using spherical microphone arrays

Human perception of room acoustics depends among others on the time of transition from early reflections to late reverberation in room impulse responses, which is known as mixing time. In this letter, a multi-channel mixing time prediction method is proposed, which in contrast to state-of-the-art channel-based predictors accounts for spatiotemporal properties of the sound field. The proposed diffuseness-based method is compared with existing model- and channel-based prediction methods through measurements and acoustic simulations, and is shown to correlate well with the perceptual mixing time. Furthermore, insights into relations between prediction methods and mixing time definitions based on reflection density are presented.