2005 Fiscal Year Final Research Report Summary
Observation and Modeling for Radiation from the Nostrils and Their Individuality
| Project/Area Number |
16500106
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Perception information processing/Intelligent robotics
|
|---|
| Research Institution | Shizuoka University |
|---|
Principal Investigator |
NAKAI Takayoshi Shizuoka University, Faculty of Engineering, Professor, 工学部, 教授 (90109132)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
TOU Takeshi Japan Advanced Institute of Science and Technology, School of Information Science, Professor, 情報科学研究科, 教授 (80334796)
|
|---|
| Project Period (FY) |
2004 – 2005
|
| Keywords | radiation from nostrils / voiced stop / buzz / velum / individuality |
|---|
| Research Abstract |
The aim of this research is to do modeling for radiation from the nostrils based on its observation and to examine its individuality. From ten male subjects, sound pressure near the nostril, vibration acceleration on the nose, and sound pressure in front of the mouth are observed using 3 channel A-D converter and ratio of radiations from the mouth and the nostril, RRMN, is observed. Speech materials consist of nasal sound /ma/ and /na/, 5 vowels, and /cvcv/(c=b,d,g, v=5 Japanese vowels/). For initial buzz, RRMN has individuality and, it is from 0dB to -15dB lower than for /m/. For medial buzz, RRMN is mostly lower than for initial buzz. Transitions from buzz to vowel and from vowel to buzz are mostly fast. From frequency analysis, it is seen that radiation from the nostrils mainly consists of fundamental frequency, the second and the third harmonics.
For three subjects in the ten male subjects, animation data of MR image are observed on mid-sagittal plane in head. From results, it is seen as follows : the thickness of the velum has individuality, it has different one due to vowels, for buzz it has different one due to the following vowels, but it has almost the same one as for the following vowels. These is different from observation of speech data, however, there is consistency in the results : the thickness of the velum is thin for /i/ and is thick for /a/, and the radiation from the nostrils is large for /i/. It is needed for observing the behavior of the velum for buzz in future work.
Speech synthesis system is constructed based on these results. When radiation is only from the nostrils for buzz and is only from the mouth for vowels, it is seen that the synthesis sound is smoother.
It is seen that for buzz radiation from the nostrils is due to laxation of the velum from described above, however, its individuality cannot be quantifiably treated.
|
Research Products
(8 results)
