Construction of Neuro-physiologically Control Model for Speech Production Process

2015 Fiscal Year Final Research Report

• Project/Area Number: 25330190
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Perceptual information processing
• Research Institution: Japan Advanced Institute of Science and Technology
• Principal Investigator: Dang Jianwu 北陸先端科学技術大学院大学, 情報科学研究科, 教授 (80334796)
• Co-Investigator(Kenkyū-buntansha): Tanaka Hirokazu 北陸先端科学技術大学院大学, 情報科学研究科, 准教授 (00332320) ; Suemitsu Atsuo 北陸先端科学技術大学院大学, 情報科学研究科, 助教 (20422199) ; Kawamoto Shin'ichi 群馬工業高等専門学校, 電子工学科, 講師 (70418507)
• Project Period (FY): 2013-04-01 – 2016-03-31
• Keywords: 音声生成 / 生理学的モデル / 発話運動制御 / 神経生理学制御

Outline of Final Research Achievements

The purpose of this study is to build up a neurophysiological control model for speech production process. Based on the common and individual characteristics of speech organs, we developed an efficient model construction approach by adapting the characteristics to a prototypical articulatory model for any given speaker, and proposed an estimation method of muscle activation patterns by reducing spatial-dependent properties of the articulatory muscles. To acquire “true” articulatory motor commands, we proposed a learning method that can reduce the effects of exiting speech category by practicing the articulatory movements alone without auditory representation beforehand. We analyzed brain activities for variety of reading words using source reconstruction method based on EEG observations. It is found that the speech processing in the brain can be explained by the dual-stream model. Based on our findings, we constructed a neurophysiological control model for speech production.

Free Research Field

音声科学、音声生成