The study to estimate the learning stage of pronunciation for hearing impaired subject from the digital video picture to speak

从数字视频图像到说话估计听障受试者发音学习阶段的研究

• 批准号: 16500620
• 负责人: AOYAMA Kensei
• 金额: $ 2.37万
• 依托单位: Tsukuba University of Technology (2005-2006)Tsukuba College of Technology (2004)
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Scientific Research (C)
• 财政年份: 2004
• 资助国家: 日本
• 起止时间: 2004 至 2006
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-16500620/
• 关键词: hearing impaired person; learing stage of pronunciation; video picture of face; three-dimensional movement analysis; 3次元動画像解析; 顔動画数; 3次元顔動画像解析

The study is to try to estimate the learning stage et pronunciation for the hearing impaired subjeru from the digital video picture to speak. The video picture to record subject putting sic makers on the lip and speaking forty words with two CCD cameras was captured in the personal computer. Every position of six markers on the lips is estimated by Move-tr/3D (the software of Ribrary Co. From the result of experiment for the normal subject, it is seen that twelve physical qualities mentioned here under are useful. (1) The abscissa of the point A (2) The component of velocity of the point A in the x direction (3) The ordinate of the point D (4) The component of velocity of the point A in the y direction (5) The length of the segment AB (distance AB) (6)The change rate of the distance AB (7) The length of the segment CD (distance CD) (8) The change rate of the distance CD (9) The angle CAD (∠CAD) (10) The change rate of the ∠CAD (11) The angle ADB (∠ADB) (12) The change rate of the ∠ADB It is seen that the extreme value of physical qualities for (1),(3),(5),(7),(9), and (11) is useful to discriminate the vowel, and the extreme value of the change rate of their qualities is useful to discriminate the consonant. As for this study, the learning stage of pronunciation for the hearing impaired subject cannot be estimated by the above mentioned qualities. But the learning stage of pronunciation will be estimated accurately by the quality for the movement of the tongue in addition to their qualities.

本研究试图从数字视频图像中估计听障人群的语音学习阶段。在个人电脑中捕捉到受试者用两个CCD摄像机将鼻塞贴在嘴唇上并说出40个单词的视频图像。用library公司的Move-tr/3D软件对嘴唇上6个标记的每一个位置进行估计。从正常受试者的实验结果来看，下面提到的12个物理品质是有用的。(1)点的横坐标的组件(2)点在x方向上的速度(3)点的纵坐标D的组件(4)点在y方向上的速度(5)段的长度AB (AB)的距离(6)距离的变化率AB(7)段的长度光盘(CD)的距离(8)距离的变化率CD(9)角度CAD(∠CAD)(10)∠CAD的变化率(11)角度亚洲开发银行(亚行∠)(12)∠亚行的变化率是见过的(1)、(3)、(5)、(7)、(9)、（11）的音质极值可用于区分元音，音质变化率极值可用于区分辅音。在本研究中，听力受损受试者的发音学习阶段不能用上述品质来衡量。但是，除了发音的质量外，还可以根据舌头的运动质量来准确地估计发音的学习阶段。