摘要
为提高识别率和识别效率,采用双门限过零率和短时能量作为端点检测的依据,提取Mel频率倒谱系数作为语音特征参数,并使用DTW算法进行模式匹配.由于传统DTW算法计算量大,所以采用局部路径约束和区域约束进行改进,并用Matlab对改进后的DTW算法进行了仿真.实验证明该算法对孤立词语音识别能够达到较好的识别结果.
To improve recognizing rate and recognizing efficiency, double-threshold zero-crossing rate is adopted in the endpoint detection, Mel-Frequency Cepstral Coefficients is obtained as speech characteristic parameters, and DTW algorithm is used for matching of the model. For the large amount of data in the matching process, DTW algorithm is improved by using local con-straints and global constraints and it was simulated by Matlab. The experiment shows that this algorithm can achieved good results in isolated word speech recognition .
出处
《山东理工大学学报（自然科学版）》
CAS
2013年第1期63-66,共4页
Journal of Shandong University of Technology:Natural Science Edition
