Behavioral Measures of Monaural Temporal Fine Structure Processing

Deficits in temporal fine structure (TFS) processing found in hearing-impaired listeners have been shown to correlate poorly to audibility and frequency selectivity, despite adverse effects on speech perception in noise. This underlines the need for an independent measure of TFS processing when characterizing hearing impairment.

Estimating the acuity of monaural TFS processing in humans however remains a challenge. One suggested measure is based on the ability of listeners to detect a pitch shift between harmonic (H) and inharmonic (I) complex tones with unresolved components (e.g. Moore et al., JASA 125:3412-3422, 2009). However, spectral cues arising from detectable excitation pattern shifts or audible combination tones might supplement TFS cues in this H/I-discrimination task.

The present study further assessed the importance of the role of TFS, in contrast to that of temporal envelope and spectral resolution, for the low pitch evoked by high-frequency complex tones. The aim was to estimate the efficiency of monaural TFS cues as a function of the stimulus center frequency Fc and its ratio N to the stimulus envelope repetition rate.

A pitch-matching paradigm was used, such that changes in spectral indices were not useable as a cue. The low pitch of broadband pulse-trains was matched to that of inharmonic transposed tones (3≤Fc≤7 kHz, N=[11.5,14.5]). Resolvability of the stimulus components was assessed, and the contribution of TFS information to individual pitch-matching results was estimated and compared to performance of the same subjects in an H/I-discrimination experiment (2.2≤Fc≤9 kHz, N=[11,13,15]) similar to that of Moore et al. (2009).

Pitch matches revealed an ambiguous low pitch related to the timing between TFS peaks near adjacent envelope maxima, up to Fc=7 kHz for N= 11.5 and Fc=5 kHz for N=14.5. Pitch salience decreased as Fc or N increased, but pitch matches never relied on the envelope repetition rate. Moreover, the results from the component-resolvability experiment indicated an inability of subjects to hear out the lowest frequency components of the stimuli.

This strongly suggests that the monaural representation of TFS persists at high frequencies and prevails over envelope or spectral cues for perception of the low pitch of high-frequency complex tones. Individual performance in the H/I-discrimination experiment is therefore expected to show a similar dependency on Fc and N as the corresponding individual pitch matching results.

If so, such methods may be useful to estimate the upper frequency limit for monaural TFS processing in individual subjects.