Multidimensional classification of normal voice qualities.
Multidimensional classification of normal voice qualities.
复制标题
正常语音质量的多维分类。
DOI：
--
复制
发表时间：
1977
影响因子：
2.4
通讯作者：
T. Murry
中科院分区：
文献类型：
--
作者：
S. Singh;T. Murry
这项研究调查并从声学角度定义了一些用于区分正常男性和女性声音的感知参数。二十名正常说话的受试者，十名男性和十名女性，制作了配对、随机的演讲片段，并呈现给十名听众评委。评委们按照七分制对所有声音组合的相似性进行评分，并通过多维分析技术 INDSCAL 来分析这些相似性。对于每个说话者，获得了八种声学测量结果，并由十名语言病理学家对四个常用描述性术语进行了心理物理学评级。这些测量和评级提供了 INDSCAL 解决方案的解释，并且还进行了相关分析。结果表明，在区分声音时，评委们最重视男性与女性的维度及其声学相关性。 INDSCAL 分析和相关分析的结果表明，听众可能使用不同的感知策略对男性声音进行分类，而不是对女性声音进行分类。
This study investigates and acoustically defines some of the perceptual parameters used to distinguish among normal male and female voices. Twenty normal speaking subjects, ten male and ten female, produced speech segments which were paired, randomized, and presented to ten listener judges. The judges rated the similarity of all combinations of voices on a seven-point scale, and these similarities were analyzed by a multidimensional analysis technique, INDSCAL. For each speaker, eight acoustical measures were obtained, and psychophysical ratings of four commonly used descriptive terms were made by ten speech pathologists. These measurements and ratings provided the interpretation of the INDSCAL solution and were also submitted to correlational analyses. The results indicate that, when distinguishing among the voices, the judges gave the greatest weight to the male versus female dimension and its acoustical correlates. The results of both the INDSCAL analysis and the correlational analyses suggest that listeners may use different perceptual strategies to classify male voices than they use to classify females ones.