Effects of rate-of-change pattern on the discrimination of unidirectional gliding tones with fixed frequency and time transition spans

Abstract

Listening experiments were designed to test the three hypotheses for detection and discrimination of glides in frequency: (1) end point sampling; (2) a weighted average method; and (3) decision based on changes in the low-frequency side of the excitation pattern. Forty-eight frequency and time transition spans were chosen. The center frequencies of testing signals were 500 Hz, 2 KHz, and 6 KHz, with frequency spans of 0, 0.5, 1, and 2 ERBs and durations of 20, 50, 200, and 400 ms. For a given transition span, the frequency difference limens for five different gliding patterns were measured, including linear chirp, quadratic and inverse quadratic sweep chirps, and FM modulated sweeps. All test patterns had the following features: (1) they are unidirectional gliding tones with increasing instantaneous frequency throughout the whole duration; (2) the phase functions are at least first-order differentiable, thus the interference of click sounds caused by fast frequency transitions are alleviated. The results showed that none of these hypotheses was able to explain the mechanism for detection and discrimination of glides in frequency well. It is speculated that, at least, both sides of the excitation patterns should be compared, and that the time-related cues could have an effect as well.