三字词声调的模糊识别方法

汉语是音节声调语言，在连续语音合成和识别中加入声调信息有重要意义。对三字词声调按首字、中字和尾字音节进行特征提取，并结合声调的特点选取相应的隶属函数进行模糊化处理，综合运用了决策树分类、模糊决策和基于知识的规则推理等技术，提出了一种模糊决策树声调识别方法，具有简便、实时性、顽健性和识别率高等特点。此方法对孤立字和二字词同样有效，并可应用于汉语连续语音。

Fuzzy Tone Recognition Method for Chinese Trisyllabic Word

Chinese is a language with syllabic tone. Tone information is very important to continuous speech recognition and synthesis. This paper extracts the feature of every syllable in trisyllabic word separately, chooses corresponding membership function to fuzzification based on the tone characteristic of Chinese trisyllabic word. After decision tree, fuzzy decision and knowledge-based rule reasoning are employed by synthesize, an approach of fuzzy decision tree for tone recognition is proposed. This method is characterized for its simpleness, real-time performance, robustness and high recognition rate. The method is available for isolated word and double-syllabic word and is applicable for Chinese continuous speech.