基于DTW的孤立词语音识别系统设计  

Design of Isolated Word Speech Recognition System Based on DTW

作　　者：叶硕 彭春堂 杜珍珍 贺娟 

机构地区：[1]武汉邮电科学研究院,湖北武汉430000

出　　处：《长江大学学报（自然科学版）》2018年第17期33-37,共5页Journal of Yangtze University(Natural Science Edition)

基　　金：湖北省科技厅2018年度湖北省技术创新专项重大项目(2018AAA063)

摘　　要：设计了一种基于小波变换的孤立词语音识别系统。该系统提取语音信号的特征矢量,使用DTW算法对待测语音进行匹配,通过对输入信号的降噪处理,提高孤立词在噪声环境下的识别精度。测试结果表明,在无强噪声干扰条件下,该系统能较好实现孤立词语音识别。Speech recognition is one of the most popular studies nowadays.This paper introduces a design of isolated word speech recognition system based on wavelet transform,it extracts the characteristic vector of speech signal,uses DTW algorithms to match the voice test.The recognition accuracy of isolated words under noise environment is improved by denoising the input signal.The experimental results show that under the condition of no strong noise interference,the system can better realize the isolated word speech recognition.

关 键 词：语音识别 小波变换 孤立词 DTW 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP391.42[自动化与计算机技术—控制科学与工程]