基于 DTW 算法的语音识别原理与实现[摘 要]以一个能识别数字 0~9 的语音识别系统的实现过程为例,阐述了基于 DTW 算法的特定人孤立词语音识别的基本原理和关键技术
其中包括对语音端点检测方法、特征参数计算方法和 DTW 算法实现的详细讨论,最后给出了在 Matlab 下的编程方法和实验结果
[关键字]语音识别;端点检测;MFCC 系数;DTW 算法[中图分类号]TN912
34[文献标识码]APrinciple and Realization of Speech Recognition Based on DTW AlgorithmAbstractWith an example of the realization of a 0~9 identifiable speech recognition system, the paper described the basic principles and key technologies of isolated word speech recognition based on DTW algorithm , including method of endpoint detection, calculation of characteristic parameters, and implementation of DTW algorithm
Programming method under Matlab andexperimental results are given at the end of the paper
Keywordspeech recognition; endpoint detection; MFCC parameter; DTW algorithm0引言自计算机诞生以来,通过语音与计算机交互一直是