Gap detection and discrimination in complex speechlike signals

Abstract

Speechlike stimuli consisting of /s/‐like noise and F 1‐like frequency sweeps were used for gap detection and discrimination experiments. Five normal‐hearing subjects participated in the tasks: (1) detection of gaps in /s/‐like noise, (2) discrimination between gaps in noise, (3) detection of gaps between noise and glide, and (4) gap discrimination between noise and glide. Stimuli for the latter two experiments were patterned after synthesized “say” and “stay” speech signals. These stimuli consisted of 120 ms of high‐pass‐filter digital white noise (simulating the /s/ phoneme), followed by a silent gap varying from 0–120 ms in duration, and a 50‐ms linear frequency sweep from 230–611 Hz (simulating the F 1 transition). The same high‐pass‐filtered white noise was used for all experiments. An adaptive psychophysical procedure was used in the first two experiments; a method of constant stimuli procedure was used for the complex stimuli tasks because of the shallow‐sloping psychometric functions. Subjects required extensive training to reach asymptotic performance levels on the latter two experimental tasks. Results indicated significantly longer durations are required for the complex stimulus tasks.