Subjective Listening Tests for the Evaluation of Generic, Mismatched, and Individualized Personal Sound Zone Filters

Abstract

Two types of subjective listening tests were conducted to evaluate and compare the perceived performance of three different personal sound zone filters: a generic filter designed using binaural room transfer functions from a dummy head, an individualized filter designed using transfer functions from the human listener, and a mismatched filter designed using transfer functions from a different human listener. The listening tests used a two-listener personal sound zone system comprised of a linear array of loudspeakers using the pressure matching method. The sound zone system models the cabin of an automobile, with the subject sitting in the driver position next to a dummy head in the passenger position. The subjective listening tests included a pink noise attenuation test which was a test using method of adjustment for the perceived attenuation of uncorrelated pink noise and a multiple stimulus hidden reference and anchor (MUSHRA) test for perceived interference for four pairs of audio programs combining music and speech signals. It was found that the individualized filter has a better perceived performance than both the generic and mismatched filters for attenuation with a mean difference of 1.40 dB for the generic filter and 1.50 dB for the mismatched filter in the pink noise attenuation test. In the MUSHRA test for interference, less interference was perceived for the individualized filter as the mean MUSHRA score for the individualized filter was 5.31 points above the mismatched filter and 3.01 points above the generic filter on a rating scale of 0-100. Additionally, it was found that the choice of audio does have an effect on perception of interference with the case in which music is played to the driver and speech is played to the passenger receiving higher overall MUSHRA scores for each filter than the other tested stimuli.