摘要
为深入研究人类与机器人的交流方式,基于语音视位建立仿人头像机器人的口形系统。考虑汉语的特点与人在发声时的口型动作,建立口形系统的基本视位库。根据基本视位库,建立仿人头像机器人的基本口形模型,采用最小二乘法识别口形系统中实际视位,并且利用Hermit插值的方法实现仿人头像机器人的动态口形。实验研究验证:该口形系统能够再现基本口形与基本视位的对应,并且动态口形实现与人类语音的协调。
In order to research on the interaction with human and robot thoroughly,the lip-shape system of humanoid head portrait robot was built based on the speech viseme. Considering the Chinese character and the human lip's movement,the basic viseme database (BVD)was built. The basic lip model was established corresponding to the basic viseme database (BVD). The least-square method was used to recognition the viseme in the BVD,dynamitic lip shape was realized by Hermit interpolation algorithm. The experimental results show that the lip shape system can represent the basic lip corresponding to basic viseme and realize the robot's dynastic lip corresponding to human speech.
出处
《机械设计与制造》
北大核心
2010年第2期152-154,共3页
Machinery Design & Manufacture
基金
国家"863计划"资助项目(2006AA04Z201)
关键词
视位
最小二乘法
仿人头像机器人
识别
Viseme
Least-Square method
Humanoid head robot
Recognition