2013 Fiscal Year Final Research Report
A research toward the development of psychological stress detecting technology using audio features.
| Project/Area Number |
24650417
|
|---|
| Research Category |
Grant-in-Aid for Challenging Exploratory Research
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
Applied health science
|
|---|
| Research Institution | University of Tsukuba |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
NEMOTO Kiyotaka 筑波大学, 医学医療系, 講師 (80550152)
|
|---|
| Project Period (FY) |
2012-04-01 – 2014-03-31
|
| Keywords | 音声分析 / 心理的ストレス / 音声パラメーター / 公開音響特徴量 |
|---|
| Research Abstract |
In this study, we aimed to develop a stress detect technology using audio features extracted from human voice. In order to achieve this aim, we set up the following three experiments using various psychological tests and voice samples; 1) a controlled study of healthy subjects under psychological tasks, 2) longitudinal observational study of healthy subjects under ordinaly lives, and 3) case control studies in order to elucidate audio features which detects depressive state. As a result, we found that psychological tasks brought a change in mel frequency cepstral coefficients and significant correlation between fundamental frequency (F0) and anxiety as well as first formant (F1) and depression. However, we were not able to detect any temporal correlation between audio features and psychological scales. We have just begun a case control studies aimed at patients with depression and keep developing a stress detection technology using voice samples.
|
