Rapid Adaptation to target and background talker variation in speech-in-speech perception

The present study examined the impact of talker variation in the target or background speech streams on speech-in-speech recognition. Listeners transcribed sentences in single-talker background babble at two signal-to-noise ratios in one of three conditions: 1) variable target talker and consistent masker talker, 2) consistent target talker and variable masker talker, or 3) consistent target and masker talkers. Results showed a significant effect of signal to-noise ratio across conditions, as well as substantial variation across target-masker pairs within conditions 1) and 2). In contrast, overall performance across all trials in all conditions was stable, suggesting that the particular energetic masking characteristics of a given target-masker talker pair within a given condition override any potential impact from talker variability in the target or in the masker across conditions. Thus, when recognizing sentence-length stimuli embedded in background speech, listeners exhibit rapid adaptation to across-trial talker variation in the target or the background.