| Project/Area Number |
02555086
|
|---|
| Research Category |
Grant-in-Aid for Developmental Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
計測・制御工学
|
|---|
| Research Institution | University of Tokyo |
|---|
Principal Investigator |
SAITO Masao Univ. of Tokyo, Faculty of Medicine, Professor, 医学部(医), 教授 (60010708)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
YAGUCHI Yoshiaki Terumo Corp., ME Department R.D. Center, researcher, 技術開発部・ME開発部門, 研究員
IKEDA Kenji Univ. of Tokyo, Faculty of Medicine, Assistant Professor, 医学部(医), 助手 (70010030)
WATANABE Akira Univ. of Tokyo, Faculty of Medicine, Associate Professor, 医学部(医), 助教授 (00009937)
|
|---|
| Project Period (FY) |
1990 – 1992
|
| Project Status |
Completed (Fiscal Year 1992)
|
| Budget Amount *help |
¥3,500,000 (Direct Cost: ¥3,500,000)
Fiscal Year 1992: ¥400,000 (Direct Cost: ¥400,000)
Fiscal Year 1991: ¥1,500,000 (Direct Cost: ¥1,500,000)
Fiscal Year 1990: ¥1,600,000 (Direct Cost: ¥1,600,000)
|
|---|
| Keywords | format transition / local peak / fricative / running spectrum / auditory filter / adaptive digital filter / スペクトルピ-ク強調 / 指向性補聴器 / ホルマント強調 / 子音強調 / 雑音抑圧 |
|---|
| Research Abstract |
We propose an adaptive digital filter which enhances transition of formants and band noise in consonants, so as to improve the articulation scores. The details are presented below.
First, for enhancement of the formant transition of consonants, we beforehand generated band-pass filters whose characteristics of frequency match the human auditory filters. Sound signals were encoded by the sampling frequency and interval which were fit for acoustic features of consonants. Spectrum loci was estimated by selective linear predictive coding in order to select the band-pass filters which cover the band width of the local peaks of spectrum. The articulation test to enhanced consonants was carried out, which became evident that the editted consonants which have enhanced formant peaks improved the articulation.
Second, a new algorithm was proposed to detect the consonants which have band noise as an acoustic feature, such as a fricative sound. Sound signals were separated into four band-pass filters, from which each amplitudes were calculated as acoustic cues for the detection. As a result, the fricative consonants were detected with three trigger outputs from the band-pass filters.
|
