Stimulus features affecting speech recognition in a two-talker masker

Abstract

Large individual differences in performance have been observed for speech recognition in the presence of two-talker speech maskers. Moreover, two-talker masker samples appear to vary in masking effectiveness. These two observations suggest that (1) listeners vary in the extent to which they take advantage of stimulus features that aid in the segregation of target from masker speech and (2) maskers vary in the extent to which salient stimulus features are available that aid in the segregation of target from masker speech. Stimulus talker variability and individual listener variability are explored across two experiments. The first experiment evaluates differences between talkers based on vocal characteristics. These characteristics include fundamental frequency, pitch variability, and cadence. Additionally, listeners are asked to rank each individual competing talker’s voice with respect to similarity to the target talker’s voice. The second experiment will explore linguistic differences in the competing speech including two-talker maskers that consist of concatenated sentences, story passages, and conversational dialogs and how the content of the speech affects masking. Data will be presented for listeners with normal-hearing thresholds.