Can temporal fine structure represent the fundamental frequency of unresolved harmonics?

Abstract

At least two modes of pitch perception exist: in one, the fundamental frequency ͑F0͒ of harmonic complex tones is estimated using the temporal fine structure ͑TFS͒ of individual low-order resolved harmonics; in the other, F0 is derived from the temporal envelope of high-order unresolved harmonics that interact in the auditory periphery. Pitch is typically more accurate in the former than in the latter mode. Another possibility is that pitch can sometimes be coded via the TFS from unresolved harmonics. A recent study supporting this third possibility ͓Moore et al. ͑2006a͒. J. Acoust. Soc. Am. 119, 480-490͔ based its conclusion on a condition where phase interaction effects ͑implying unresolved harmonics͒ accompanied accurate F0 discrimination ͑implying TFS processing͒. The present study tests whether these results were influenced by audible distortion products. Experiment 1 replicated the original results, obtained using a low-level background noise. However, experiments 2-4 found no evidence for the use of TFS cues with unresolved harmonics when the background noise level was raised, or the stimulus level was lowered, to render distortion inaudible. Experiment 5 measured the presence and phase dependence of audible distortion products. The results provide no evidence that TFS cues are used to code the F0 of unresolved harmonics.

Does fundamental-frequency discrimination measure virtual pitch discrimination?

“…Had we simply computed geometric means of the DLCs from the two types of track, we would have obtained a rather different result: nearly identical thresholds for inharmonic complexes ͑geometric mean across feedback and no-feedback conditions= 0.33͒ and inharmonic complexes ͑geometric mean across feedback and no-feedback conditions= 0.34͒. Most other studies that have roved the lowest harmonic number of a complex tone ͑e.g., Houtsma and Smurzynski, 1990;Bernstein and Oxenham, 2003;Oxenham et al, 2009͒ have roved randomly within a single track, making our comparisons with earlier data dependent on the assumption that Eq. ͑1͒ is a valid way of estimating DLCs in this context.…”

“…Moreover, the findings confirm and quantify the influence of randomizing the lowest harmonic number across intervals ͑within trials͒, and they indicate that although perhaps not absolutely necessary, when applied properly, such randomization reduces the influence of timbre or other cues related to changes in the frequency of the lowest harmonic-but unrelated to residue-pitch perception. Evidently, the roving technique must be adequate to ensure that the level of performance targeted or measured in an experiment cannot be accounted for based on residual cues related to the frequency of the lowest harmonic ͑see Oxenham et al, 2009 for a discussion of the merits and limitations of different randomization techniques͒. The roving technique that was used in the current study, in combination with the dualtracking procedure and the method for calculating unbiased thresholds, provides a useful approach for the measurement of difference limens for residue pitch, while simultaneously assessing the influence of cues related to the frequency of the lowest harmonic, including timbre ͑brightness͒ differences.…”

Relating binaural pitch perception to the individual listener’s auditory profile

“…Uncertainties persist concerning the role of a temporal mechanism for FM detection at very low rates (Lacher-Fougère and Demany, 1998), the measure chosen in the present study. Another disputed method is the discrimination of harmonic and frequency-shifted bandpass-filtered complex tones (Moore and Sęk, 2009), for which the sole role of TFS information has been questioned (Oxenham et al, 2009), and which cannot be used at low frequencies. The search for a psychophysical or physiological outcome that would accurately reflect peripheral TFS-processing abilities thus ought to be pursued.…”

Development of a method for determining binaural sensitivity to temporal fine structure

“…Firstly, the H and I tones do have slightly different excitation patterns, and in principle these differences could be used to perform the task (Micheyl, Schrater, and Oxenham 2013), especially when combination tones are taken into account (Oxenham, Micheyl, and Keebler 2009), although there are several lines of evidence suggesting that the task is not performed using excitation-pattern cues Sek 2009, 2011;Jackson and Moore 2014;Marmel et al 2015). Secondly, some older people with hearing loss (Hopkins and Moore 2007) and with normal audiograms (Moore, Vickers, and Mehta 2012;Füllgrabe, Moore, and Stone 2015) cannot perform the TFS1 test at all, so the test cannot be used to obtain a graded measure of sensitivity to TFS in those listeners.…”