论文部分内容阅读
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。
Chinese is a syllable tonal language. It is of great significance to add tonal information to continuous speech synthesis and recognition. In the first word, middle word and tail word syllable, the feature extraction of the three-word tone is carried out, and the corresponding subordinate function is selected according to the characteristics of the tone to carry out the fuzzification processing. The technology of decision tree classification, fuzzy decision and rule-based reasoning based on knowledge , A fuzzy decision tree tone identification method is proposed, which has the characteristics of simplicity, real-time, robustness and high recognition rate. This method is equally valid for isolated words and two words, and can be applied to Chinese continuous speech.