Stassen H H
Psychiatric University Hospital Zurich, Research Department, Switzerland.
Psychopathology. 1988;21(2-3):83-8. doi: 10.1159/000284547.
It is well known that the human voice contains important information about the affective state of a speaker at a nonverbal level. Accordingly, we started an extensive investigation which aims at modelling intraindividual changes of the global affective state over time, as this state is reflected by the human voice, and can be inferred from measurable speech parameters. For the purpose of this investigation, a speech-recording procedure was designed which is especially suited to reveal intraindividual changes of voice patterns over time since each person serves as his or her own reference. On the other hand, the chosen experimental setup is less suited to classify patients in the sense of a traditional diagnostic scheme. In order to find an appropriate mathematical model on the basis of speech parameters, a calibration study with 190 healthy subjects was carried out which enabled us to investigate each parameter for its reproducibility, sensitivity and specificity. In particular, this calibration study yielded the information of how to draw the line between 'normal' fluctuations and 'significant' intraindividual changes over time. All speech parameters under discussion turned out to be sufficiently stable over time, whereas, in regard to their sensitivity to form and content of text, significant differences showed up. In a second step, a pilot study with 6 depressive patients was carried out in order to investigate the specificity of voice parameters with regard to psychopathology. It turned out that the registration procedure is realizable even if patients are considerably handicapped by their illness. However, no consistent correlations could be revealed between single speech parameters and psychopathological rating scales.(ABSTRACT TRUNCATED AT 250 WORDS)
众所周知，人类的声音在非语言层面包含了有关说话者情感状态的重要信息。因此，我们展开了一项广泛的调查，旨在对全球情感状态随时间的个体内变化进行建模，因为这种状态由人类声音反映出来，并且可以从可测量的语音参数中推断出来。为了此次调查的目的，设计了一种语音记录程序，该程序特别适合揭示语音模式随时间的个体内变化，因为每个人都以自己作为参照。另一方面，所选的实验设置不太适合按照传统诊断方案对患者进行分类。为了在语音参数的基础上找到合适的数学模型，对190名健康受试者进行了一项校准研究，这使我们能够研究每个参数的可重复性、敏感性和特异性。特别是，这项校准研究得出了如何区分“正常”波动和随时间的“显著”个体内变化的信息。所有讨论中的语音参数在一段时间内都表现出足够的稳定性，然而，就它们对文本形式和内容的敏感性而言，出现了显著差异。第二步，对6名抑郁症患者进行了一项初步研究，以调查语音参数在精神病理学方面的特异性。结果表明，即使患者因疾病而严重受限，记录程序也是可行的。然而，单一语音参数与精神病理学评定量表之间未能揭示出一致的相关性。（摘要截断于250字）
