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%。 Discussed the method of voiced / unvoiced classification of speech, and focused on the speech classification system based on multiple characteristic parameters and applying pattern recognition theory. Experiments show that the classification method based on Fisher classifier can classify the accuracy of the speech within the training up to 98.6%, and the accuracy of the classification of the speech outside the training can reach 96.7%.