基于DTW语音识别的研究与实现
首发时间:2010-04-01
摘要:本文描述了基于DTW模型语音算法的优化与改进并对其进行了实现。对端点检测的方法作了改进,将可变窗长的语音端点检测与双门限端点检测方法结合起来,同时在DTW语音识别的过程中采用了松弛起点与终点的办法,将改进后的仿真结果与传统的语音算法仿真结果作了比较,对语音识别作到了优化处理。
For information in English, please click here
The Research And Implementation Based On The Improvement Of The DTW Speech Recognition
Abstract:This paper describes the model based on the DTW algorithm for optimization and improvement of speech and the realization of them. It has been improved the endpoint detection method. The variable-length windows and double-Speech Endpoint Detection threshold combined the endpoint detection methods. While the relaxation approach used in the process of starting and ending of DTW speech recognition, the results will be improved with by the algorithm. Compared with the traditional voice-simulation results, it is optimized in the voice recognition.
Keywords: Speech Recognition DTW Model Endpoint Detection Dynamic Warping
基金:
论文图表:
引用
No.4142952003112701****
同行评议
共计0人参与
勘误表
基于DTW语音识别的研究与实现
评论
全部评论