Effect of relative masker levels on speech recognition in two-talker maskers with varying perceptual similarity to the target speech

Abstract

Speech recognition thresholds were measured as a function of the relative level between two speech maskers that differed in perceptual similarity from the target. Results showed that recognition thresholds were driven by the relative level between the target and perceptually similar masker when the perceptually similar masker was softer, and by the relative level between the target and both maskers when the perceptually similar masker was louder. This suggests that effectiveness of a two-talker masker is primarily determined by the masker stream that is most perceptually similar to the target, but also by the relative levels between the two maskers.