Wolfe V, Cornell R, Palmer C
Speech and Hearing Clinic, Auburn University, Montgomery, AL 36193-0408.
J Speech Hear Res. 1991 Jun;34(3):509-16. doi: 10.1044/jshr.3403.509.
Listeners classified 49 samples of vowels /a/ and /i/ on the basis of four voice types: hoarse, breathy, strained, and normal. The vowels were analyzed acoustically for mean harmonic/noise differences in four spectral regions, average fundamental frequency, natural logarithm of fundamental frequency, and jitter. Discriminant analysis showed that classifications of voice type were made with 80% accuracy using three acoustic parameters: (a) mean harmonic/noise difference factor (1-3.5 kHz), (b) natural log of fundamental frequency, and (c) vowel type. The significance of these particular acoustic parameters for the perception and classification of voice types is discussed.
嘶哑型、气息型、紧张型和正常型，对49个/a/和/i/元音样本进行了分类。对这些元音进行了声学分析，以确定四个频谱区域的平均谐波/噪声差异、平均基频、基频的自然对数和抖动。判别分析表明，使用三个声学参数可以80%的准确率对嗓音类型进行分类：(a)平均谐波/噪声差异因子(1 - 3.5千赫)，(b)基频的自然对数，以及(c)元音类型。讨论了这些特定声学参数对于嗓音类型感知和分类的重要性。
