Pilot study on the perceptual quality of close-mic recordings in auralization of a string ensemble

Perceptually convincing auralization of joint musical performances requires an accurate representation of individual source signals. Recording instruments in anechoic environments typically lack the natural and intrinsic characteristics of joint performances due to limited room acoustic and inter-musician feedback. An alternative is to use close miking techniques to capture individual instruments in a joint performance, but minimizing microphone cross-talk and room acoustic feedback is a challenge here. This study investigates the perceptual quality of close-mic recordings in auralization of an ensemble in comparison to the binaural recordings. Binaural and close-mic recordings of a string ensemble performance with different number of violins were collected from Detmold concert house, and its auralization was performed using the Binaural Room Impulse Responses (BRIRs) obtained from in-situ measurements and room acoustic simulation. The recorded and auralized samples were perceptually evaluated by expert listeners in terms of their naturalness and similarity. Although the binaural recordings were not always rated to be highly natural, the clip-on microphone signals auralized using BRIR measurements show a similar distribution of naturalness to them which demonstrates their applicability for auralization-related applications. However, less natural and dissimilar ratings of simulated samples indicate the need for improvement of auralization using geometry-based room acoustic simulations.