RESEARCH ON THE SPEECH ANALYSES OF BABIES' CRIES FOR THE RECOGNITION OF THEIR MENTAL STATES
| Project/Area Number |
15560337
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Communication/Network engineering
|
|---|
| Research Institution | MEIJI UNIVERSITY |
|---|
Principal Investigator |
ARAKAWA Kaoru MEIJI UNIVERSITY, SCHOOL OF SCIENCE AND TECHNOLOGY, PROFESSOR, 理工学部, 教授 (30183734)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
ITOH Setsuko DOSHISHA WOMEN'S COLLEGE, DEPT.OF HUMAN LIFE STUDIES, 生活科学部, 教授 (50144358)
|
|---|
| Project Period (FY) |
2003 – 2004
|
| Project Status |
Completed (Fiscal Year 2004)
|
| Budget Amount *help |
¥3,700,000 (Direct Cost: ¥3,700,000)
Fiscal Year 2004: ¥1,400,000 (Direct Cost: ¥1,400,000)
Fiscal Year 2003: ¥2,300,000 (Direct Cost: ¥2,300,000)
|
|---|
| Keywords | baby cries / speech analysis / babies and infants / signal processing / childcare / state estimation / nonlinear filter / noise reduction |
|---|
| Research Abstract |
A system for recognizing mental states of babies is proposed from the frequency analyses of their cries. In the previous research, the researchers clarified that the babies' cries can be classified into hungry ones and sleepy ones by the frequency analyses, if the babies are elder than or equal to 5 months old. Since the frequency distribution of cries differs depending on their month age, the same method for classification cannot be applied to the cries of younger babies, under 5 months old.
In this research, another difference was found in the frequency distribution of hungry cries and sleepy ones for babies aged from 3 to 4months old. Crying voices were collected from 16 babies, both male and female, and the rough forms of the amplitude spectra were examined. For hungry cries, one of the following tendencies was observed.
1)The spectrum falls steeply around 6kHz.
2)The spectrum keeps a high value till about 8-9kHz and falls gradually.
3)The spectrum falls gradually from 0 kHz and takes
… More
a convex form upward.
Otherwise, babies were often in a sleepy state. Using these tendencies, a rule-based system which classifies a hungry state and a sleepy one was proposed for babies aged from 3 to 4 months old. About 80% of the cries of the 16 babies were correctly classified into hungry and sleepy states with this system. Some classification results are judged as mistakes, because babies are sometimes in both hungry and sleepy states and it is difficult to obtain the true reason for the cry. Also the correct classification rate depends on babies a little. The system was shown to be valid for most of the 16 babies, but not for one of them.
Since noise reduction of the babies' cries is necessary for this research, a nonlinear digital filter for noise reduction of speech signal was proposed by the head investigator before. New research on the application of this nonlinear digital filter to noise reduction of image and the detection of watermark signal in digital cinema was also performed. Less
|
Report
(3 results)
Research Products
(19 results)
