1997 Fiscal Year Final Research Report Summary
Study on Speaker Individuality in Speech and its Control
| Project/Area Number |
07680388
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | Japan Advanced Institute of Science and Technology |
|---|
Principal Investigator |
AKAGI Masato Japan Advanced Institute of Science and Technology, School of Information Science, Associate Professor, 情報科学研究科, 助教授 (20242571)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
IWAKI Mamoru Japan Advanced Institute of Science and Technology, School of Information Scienc, 情報科学研究科, 助手 (20262595)
IIJIMA Taizo Japan Advanced Institute of Science and Technology, School of Information Scienc, 情報科学研究科(7年度縲怩W年度), 教授 (50016622)
|
|---|
| Project Period (FY) |
1995 – 1997
|
| Keywords | speaker individuality / physical correlates of speaker individuality / vocal tract / spectral envelope / vocal chord / fundamental frequency / lateral misarticulation |
|---|
| Research Abstract |
Although speech can be described by its physical aspects ; that is fine spectral envelopes reflecting vocal tract features and fundamental frequencies related to glottal vibration characteristics, the physical correlates of speaker individuality embedded in the physical aspects have not been discussed in detail. This report investigates speaker individuality in spectral envelopes and fundamental frequency contours. Additionally, this report focuses on speech timbre related to spectral envelope shapes through acoustical features and articulatory movement of lateral misarticulations (LM). The findings lead to the following conclusions.
1. Spectral envelope : (1) the peaks in the spectral envelopes are more significant than dips for speaker identification, (2) speaker individuality mainly exists in the frequency band higher than the peak around 1740 Hz, and (3) the voice quality can be controlled by replacing the frequency band of one speaker with that of other speakers.
2. Fundamental frequency : (1) Speaker individuality exists in the fundamental frequency contours, (2) some specific parameters related to the height and the timing of the fundamental frequency have more speaker individuality features, and (3) the manipulation of these parameters can control speaker individuality.
3. Lateral misarticulation : (1) the spectral envelopes of LM are flat in the frequency band above approximately 4 kHz and there is a substantial peak around 3.2 kHz in LM,which varies with time almost periodically ; (2) the replacement of the spectral envelope between 2.5 and 4.5kHz of normal speech with that of LM resulted in a remarkable increase auditory impressions of LM,and (3) the typical peaks of LM are related to the length and the position of the vocal tract constriction region.
|
