- 著者
-
粕谷 英樹
鈴木 久喜
城戸 健一
- 出版者
- 一般社団法人日本音響学会
- 雑誌
- 日本音響学会誌 (ISSN:03694232)
- 巻号頁・発行日
- vol.24, no.6, pp.355-364, 1968-11-30
- 被引用文献数
-
43
For the purpose of determining the characteristics of phonetic quality of voewls, the trend in the changes of formants and pitch of vowel with age and sex of speaker has been investigated by analyzing a large number of samples of five japanese vowels. Conclusions obtained are as follows. (1) Most of the formants considerably change with the age of the speaker. Principal articulatory factor in the change of formants is the difference in the vocal tract length. On the other hand, the third formant of /i/ which depends mainly upon the front part of oral cavity, and the first and second formants of vowels /o/ and /u/ which constitute comparatively lip-rounded articulatory configuration do not change so much. For each case of children, youth and female adults the ratio of the measured formant of the open and back vowel /a/ to the mean value of the corresponding formant of male adults is approximately constant. Hence we can estimate the vocal tract length of the speaker from the mesured formants, using Eq. (2). (2) The differences between the first and second formants of male and those of female become distinct after 11 years old, while the difference between the third formant of male and that of female becomes distinct after 9 years old. absolute differences of formants, particularly of the third formant, are useful to discriminate the sex of the speaker. This is important in the case of children since pitch is useless in the distinction of the sex of the speaker before 12 years old (the voice change). (3)There is obvious difference between the pitches of children, youth, female adults and male adults, but it is difficult to infer the age of the speaker from his pitch. (4) Generally speaking, there is a correlation between formant and pitch, but there is no correlation if the speech samples are taken from the speakers of the same age. The correlations of formants and pitch come from the correlations between the age and the formants and the correlation between the age and the pitch. (5) Perfect discrimination of the vowels can not be made by the first and second formants only. There are some confusions between certain vowels (/a/ and /o/, /e/ and /u/) on the first and second formant-plane. But, thre is little confusion between the vowels in the three dimentional space composed of the first, the second and the third formants or of the pitch, the first and the second formants. The pitch or the third formant, not to mention the first and second formant, is an indispensable parameter for the discrimination of the vowels regardless of the age and sex of the speaker.