Discrimination of missing fundamental frequency and the influence of competing pitch and timbre cues

Abstract

The DL for frequency discrimination of complex tones has generally been shown to be smaller than for pure tones of equivalent F0. A notable exception is the case where “residue” tones are used [A. Faulkner, J. Acoust. Soc. Am. 78, 1993–2004 (1985)]. It has been argued that the decline in performance is due to differences in timbre that make pitch discrimination difficult. The present experiment atttempts to determine such perceptual correlates of changes in fundamental frequency for complex tones presented in two‐tone sequences. All tones comprised four equal‐amplitude harmonics m, (m + 1), (m + 2), and (m + 3), with m = 1, 2, 3, 4, or 5. The first tone in a sequence had F0 = 200 Hz, while the second had F0 = (200 + 2n) Hz, where n = 0, 1, 2, 4, 8, 16, or 32. The spectral locus of tones in a pair was either the same (m,m) or different [(2,m) or (m,2)], where m and 2 denote the harmonic number of the lowest component. Listeners were required to indicate if the tones in a pair were: (1) the same, (2) different in pitch, (3) same in pitch, but different in “something else,” or (4) different both in pitch and “something else.” Results indicate that listeners use labels 1 and 2 for sequences in which tones had the same locus, with relative use of labels systematically varying across differing magnitudes of F0 change. When spectral locus changes, however, a considerable variation of responses is seen, dependent on overall spectral proximity as well as amount of change in F0. The order of locus change across tones, whether higher or lower in frequency, is another factor that influenced label choice. [Work supported by AFOSR.]