Differences in frequency modulation detection and fundamental frequency discrimination between complex tones consisting of resolved and unresolved harmonics

Publisher Website
Google Scholar
Abstract
Three experiments investigated the finding [Shackleton and Carlyon, J. Acoust. Soc. Am. 95, 3529–3540 (1994)] that listeners are very poor at detecting frequency modulation(FM) imposed on complex‐tone carriers consisting of harmonics unresolved by the peripheral auditory system. In the first experiment, stimuli consisted of the harmonics of a 62.5‐ or 250‐Hz fundamental (f 0) passed through a bandpass filter with cutoff frequencies of 1375 and 1875 Hz. This produced two complexes, one with resolved and one with unresolved harmonics. Threshold measurements confirmed that listeners were very poor at detecting 5‐Hz FM imposed on a 200‐ms unresolved group, but that performance was relatively good for the resolved group. In addition, the results revealed that, for the unresolved group, thresholds for detecting FM were similar to thresholds for detecting an f 0 difference between two static 50‐ms complex‐tone bursts presented 50 ms apart. The second experiment added a third complex consisting of the harmonics of a 250‐Hz f 0 passed through a bandpass filter with cutoff frequencies of 5500 and 7500 Hz. These harmonics were also unresolved. Both groups of unresolved harmonics produced similar results, indicating that f 0 per se did not determine the results from the first experiment. The second experiment confirmed the results of the first experiment and in addition showed that, for unresolved harmonics only, there was a large increase in the frequency difference limen (FDL) for changes in f 0 as the duration of the tone was decreased from 200 to 50 ms. A final experiment was conducted to examine this duration effect. A fourth complex was added which included various harmonics of a 250‐Hz f 0 chosen so that no two harmonics were the same between the two observation intervals. This complex was low‐pass filtered at 1875 Hz. For this second group of resolved harmonics, listeners could not perform the FDL task by comparing the frequencies of individual harmonics: some computation of f 0 was required. FDLs for the two resolved groups showed little variation with duration whereas the FDLs for the unresolved groups increased dramatically as the duration was decreased below 100 ms. Measurement of psychometric functions showed that this result reflected a differential effect of duration on detectability itself, as characterized by d′. It is argued that the poor performance on FM detection for groups of unresolved harmonics reflects a poor performance on f 0 extraction over short durations. Furthermore, it is suggested that the differential effects of duration for the resolved and unresolved harmonics may reflect the operation of two distinct pitch mechanisms.

This publication has 0 references indexed in Scilit: