摘要
双音节二字词的声调模式比孤立字四声模式复杂得多,本文提出一种非特定人二字词声调模糊识别方法.用基音曲线的主要特征构成模糊隶属函数.声调识别算法中用决策函数进行声调模式分类,方法具有简便、可靠、实时性和顽健性好等特点.实验的识别率为96.8%.
The double-syllabic word of Chinese speech is composed of two single syllables, but its tone pattern is more complicated than that of isolated word. In this paper a speaker independent fuzzy tone recognition approach for the double-syllabic word is proposed. The main features of pitch contour are used to construct, the fuzzy menbership functions. The tone recognition algorithm ueses the decision function for tone pattern classification. This methed is characterized for its simple, reliable, real-time performance and robustness. The recognition rate is about 96.8% in our experiments.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
1998年第1期82-88,共7页
Pattern Recognition and Artificial Intelligence
基金
国家自然科学基金
关键词
二字词
声调识别
模糊相似度
语音识别
模式识别
Double-Syllabic Word, Pitch Detection, Tone Recognition, Fuzzy Similarity
