摘要
动态时间规整(Dynamic Time Warping,DTW)是语音识别的一种简单有效的方法,该算法基于动态规划的思想,解决了发音长短不一的模板匹配问题,是语音识别中出现较早、较为经典的一种算法。这里改进了传统的DTW算法,将其应用到实时语音识别系统中,并在计算机上进行了仿真。实验结果表明,改进后的算法,能有效提高孤立词的识别性能。
Dynamic time warping（Dynamic Time Warping,DTW） is a simple and effective speech recognition method.The algorithm is based on dynamic programming to solve the pronunciation of different lengths of the template matching problem,which is the speech recognition appeared earlier,more classical an algorithm.This study improved the traditional DTW algorithm,applied it into real-time speech recognition systems and into computer simulation.Experimental results showed that the improved algorithm can effectively improve the performance of isolated word recognition.
出处
《中国西部科技》
2011年第1期38-39,94,共3页
Science and Technology of West China
关键词
语音识别
动态时间规整
模板匹配
孤立词
Speech recognition
Dynamic time warping
Template matching
Isolated word
