| 基于DSP的语音识别系统研究 |
| |
| 引用本文： | 翟片富,景新幸.基于DSP的语音识别系统研究[J].大众科技,2013(12):16-18. |
| |
| 作者姓名： | 翟片富 景新幸 |
| |
| 作者单位： | 桂林电子科技大学信息与通信学院,广西桂林,541004 |
| |
| 摘 要： | 文章介绍了语音识别的基本原理以及用DSK6713实现语音识别算法的一些原则和方法，阐述了语音识别在DSP上的实现技术。系统使用梅尔倒谱系数（MFCC）作为特征参数，采用算法相对简单以及计算量较小的动态时间弯折算法（DTW）实现语音参数的匹配。用MATLAB实现DTW算法的仿真，进而将语音识别技术应用到DSP上，实验结果表明对特定人、小词汇量和孤立词的语音识别效果比较好。
|
| 关 键 词： | 语音识别 DSP 梅尔倒谱特征参数 DTW |
Study of speech recognition system based on DSP |
| |
| Abstract: | This paper describes the basic principles of speech recognition and DSK6713 implemented some principles and methods of speech recognition algorithm, explain the DSP realization technology of speech recognition. The system of speech recognition adopts Mel Frequency Cepstrum Coefficient （MFCC） to catch speech characteristic parameters and introduce dynamic time wrpping （DTW） arithmetic to realize speech pattern matching. The algorithm was simulated by Matlab software, and applies them into DSP, and the experiment results show that there has well effect on the specific person, ioslated words and small ocabulary speech recognition system. |
| |
| Keywords: | Speech recognition DSP Mel Frequency Cepstrum Coefficient DTW |
| 本文献已被 维普 等数据库收录！ |
