基于模式识别的语音分类方法  

A Pattern Recognition Decision Method for the Voiced/Unvoiced Classification of Speech

作　　者：朱琦[1] 酆广增[1] 肖海勇[1] 

机构地区：[1]南京邮电学院通信工程系,江苏南京210003

出　　处：《南京邮电学院学报》2000年第4期29-33,共5页Journal of Nanjing University of Posts and Telecommunications(Natural Science)

摘　　要：对语音的浊 /清分类方法进行了讨论 ,并着重研究了基于多个特征参数、应用模式识别理论的语音分类系统。实验表明 ,采用Fisher分类器实现的语音分类方法 ,对训练内语音分类准确率可达98 6 % ,而训练外语音分类准确率也可达 96 7%。Some methods for the classification of voiced/unvoiced speech are explored in this paper,and the focus is on the decision-making process,which can be viewed as a pattern recognition problem.Experiments demonstrate that the classifier based on the Fisher rule can achieve 98.6 percent classification accuracy on speaker dependent tests,and 96.7 percent accuracy on speaker independent tests.

关 键 词：语音数据处理 模式识别 语音分类 

分 类 号：TN912.3[电子电信—通信与信息系统]