摘要
提出了一种基于ＢＰ人工神经元网络的汉语语音声韵母分割算法．计算机模拟结果表明，该算法只需对极少数典型音节进行简单训练，便可实现汉语语音的声韵母分割，且分割精度远大于传统算法所能获得的精度．
A new approach to Chinese Shengmu and Yunmu detection based on the BP algorithm in a multiplayer perceptron (MLP) is proposed. The results of computer simulation show that the detection of Chinese Shengmu and Yunmu can be achieved with a simple training of only a few typical Chinese syllables. The accuracy is far better than that by the traditional approach. The detection results concerning the numerals (0￣9) in Chinese are given. The effectiveness is verified and the experimental results are discussed.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
1996年第S2期30-32,共3页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
中国博士后和武汉青年晨光计划基金
关键词
神经网络
模式识别
语音信号处理
neural network
pattern recognition
speech signal processing
