Moore B C, Glasberg B R
Department of Experimental Psychology, University of Cambridge, England.
J Acoust Soc Am. 1989 Oct;86(4):1351-65. doi: 10.1121/1.398695.
These experiments measure the ability to detect a change in the relative phase of a single component in a harmonic complex tone. Complex tones containing the first 20 harmonics of 50, 100, or 200 Hz, all at equal amplitude, were used. All of the harmonics except one started in cosine phase. The remaining harmonic started in cosine phase, but was shifted in phase half-way through either the first or the second of the two stimuli comprising a trial. The subject had to identify the stimulus containing the phase-shifted component. For normally hearing subjects tested at a level of 70 dB SPL per component, thresholds for detecting the phase shift [i.e., phase difference limens (DLs)] were smallest (2 degrees-4 degrees) for harmonics above the eighth and for the lowest fundamental frequency (F0). Changes in phase were not detectable for harmonic numbers below three or four at the lowest F0 and below 5-13 at the highest F0. The DLs increased slightly for the highest harmonics in the complexes. The DLs increased markedly with decreasing level, except for the highest harmonic, where only a small effect of level was found. Subjects reported that the phase-shifted harmonic appeared to "pop out" and was heard with a pure-tone quality. A pitch-matching experiment demonstrated that the pitch of this tone corresponded to the frequency of the phase-shifted component. For the highest harmonic, the phase shift was associated with a downward shift of the edge pitch heard in the reference (all cosine phase) stimulus. When the phases of the components in the reference stimulus were randomized, phase DLs were much higher (and often impossible to measure), the pop-out phenomenon was not observed, and no edge pitch was heard. Subjects with unilateral cochlear hearing impairment generally showed poorer phase sensitivity in their impaired than in their normal ears, when the two ears were compared at equal sound-pressure levels. However, at comparable sensation levels, the impaired ears sometimes showed lower phase DLs. The results are explained by considering the waveforms that would occur at the outputs of the auditory filters in response to these stimuli.
这些实验测量了在谐波复合音中检测单个成分相对相位变化的能力。使用了包含50、100或200赫兹的前20个谐波且所有谐波幅度相等的复合音。除了一个谐波外，所有谐波都以余弦相位开始。其余谐波以余弦相位开始，但在构成一次试验的两个刺激中的第一个或第二个刺激的中途相位发生了偏移。受试者必须识别出包含相位偏移成分的刺激。对于在每个成分70分贝声压级下测试的正常听力受试者，对于高于第八个谐波以及最低基频（F0），检测相位偏移的阈值[即相位差阈（DLs）]最小（2度 - 4度）。对于最低F0时低于三或四个谐波以及最高F0时低于5 - 13个谐波，相位变化无法检测到。复合音中最高谐波的DLs略有增加。除了最高谐波，DLs随着声压级降低而显著增加，最高谐波仅发现声压级有小的影响。受试者报告说，相位偏移的谐波似乎“突显出来”，并且听起来具有纯音质量。一个音高匹配实验表明，这个音的音高与相位偏移成分的频率相对应。对于最高谐波，相位偏移与在参考（全余弦相位）刺激中听到的边缘音高的向下偏移相关。当参考刺激中成分的相位随机化时，相位DLs要高得多（且通常无法测量），未观察到突显现象，也听不到边缘音高。当在相等声压级下比较双耳时，单侧耳蜗听力受损的受试者在其受损耳中的相位敏感性通常比正常耳差。然而，在可比的感觉水平下，受损耳有时显示出较低的相位DLs。通过考虑响应这些刺激时听觉滤波器输出处会出现的波形来解释这些结果。
