Intelligibility of Target Signals in Sequential and Simultaneous Segregation Tasks

: Two experiments are described in the report: the first experiment investigated target intelligibility in a sequential segregation task, while the second experiment investigated performance in a simultaneous segregation task. In the first experiment, speech intelligibility was examined in the presence of four types of maskers (continuous noise, interrupted noise, continuous speech and interrupted speech) at seven interruption rates. Results highlight the important role that target-masker similarity plays in the segregation of sequentially-presented speech signals. In the second experiment, we assessed whether listeners could extract information from target phrases that were masked either by contextually similar maskers that were confusable with the target or contextually irrelevant maskers where no such confusion was possible. The results show that a large multimasker penalty occurs when at least one of the interfering talkers is contextually similar to the target, but that no multimasker penalty occurs when neither of the masking voices is confusable with the target. These results suggest that the primary impact of the multimasker penalty is a severe degradation in the listener's ability to link together words that occur at different points in the target phrase.