基于自组织神经网络的数字语音识别  

Digital Speech Recognition Based on Self-organizing Neural Network

作　　者：曹鹏[1] 郑雪燕[1] 徐伟[1] 

机构地区：[1]空军工程大学导弹学院,陕西三原211400

出　　处：《电声技术》2002年第4期10-12,共3页Audio Engineering

摘　　要：提出一种基于自组织神经网络的数字语音识别模型。首先用基于小波变换和线性预测的特征提取方法提取语音信号特征,用自组织神经网络进行识别判决。这种语音识别方法适合于小词汇量的孤立词识别,网络结构简单,所需训练数据十分的少,实时性能好。用MATLAB进行仿真实验,识别率达到98%。The Speech Recognition method is presented,which is based on the Self-Orga-nizing Neural Networks.This method is good at small vocabulary isolated word recognition.It is simple and speedy.It needs a few data to train the network.The experimental result shows that the recognition accuracy is better than98%.[

关 键 词：自组织神经网络 数字语音识别 语音信号特征提取 

分 类 号：TN912.34[电子电信—通信与信息系统]