Psychometric functions of sentence recognition in amplitude-modulated noises

Abstract

Although it is known that listeners can take advantage of the fluctuations in the temporal envelopes of background noises to improve the recognition of target messages, the amount of benefit is difficult to compare across listeners and experimental conditions. One solution to this difficulty is to estimate the psychometric functions for speech recognition instead of the speech reception threshold alone (i.e., the 50% correct point on the psychometric function). The current study utilized a rapid psychophysical procedure that enabled the robust estimation of the psychometric function for sentence recognition in noise using as few as 20 sentences. Using this procedure, sentence recognition was measured by presenting target sentences in amplitude-modulated noise maskers. In separate conditions, the target intensity (40 or 70 dB SPL) and the masker modulation rate (1–64 Hz) were systematically varied. Manipulating these two stimulus parameters influenced both the speech reception threshold and the slope of the psychometric function. Data collected from ten young, normal hearing listeners indicated that the fluctuating-masker benefit was much more evident at the higher target level and it exhibited non-monotonic dependencies on masker modulation rate.