论文部分内容阅读
借助汉语发音口形的生理特点，在音素识别这一水平上进行汉语语音的辅助识别，具体给出了一种口形形状识别和灰度的统计方法及其具体实现．实验结果基本与理论估算相吻合，对５个元音的口形区别正确率在８０％以上，为语言的声波识别提供了一种有利的辅助手段
With the help of the physiological characteristics of Chinese articulation, aided recognition of Chinese speech at the level of phoneme recognition, a statistical method of lip shape recognition and grayscale and its implementation are given. The experimental results are in good agreement with the theoretical estimates. The accuracy of the five vowels’ mouth shapes is above 80%, which provides a favorable aids for the sonic identification of speech