| 结合残差相位的MFCC特征改进算法 |
| |
| 引用本文： | 俸云,景新幸.结合残差相位的MFCC特征改进算法[J].计算机仿真,2009,26(10):327-329,343. |
| |
| 作者姓名： | 俸云 景新幸 |
| |
| 作者单位： | 桂林电子科技大学信息与通信学院; |
| |
| 基金项目： | 面上项目(Z20656) |
| |
| 摘 要： | 美尔频率倒谱参数(Mel frequency cepstral coefficient,MFCC)仿真了人耳的听觉特性,在语音识别实际应用中取得了比较高的识别率。为了更进一步完善系统以提高系统的识别率,提出一种将MFCC和残差相位相结合的方法进行语音识别。将传统的基于MFCC的语音识别效果,与基于MFCC和残差相位相结合的语音识别效果进行比较。通过在MATLAB环境下进行仿真实验得出理想结论。利用MFCC和残差相位相结合的识别率高于MFCC的系统的识别率。所提出的改进算法更好的完善了识别系统,获得了更高的语音识别率。
|
| 关 键 词： | 语音识别 美尔频率倒谱系数 残差相位 识别率 |
Combining Residual Phase for Improving MFCC Features |
| |
| FENG Yun,JING Xin-xing.Combining Residual Phase for Improving MFCC Features[J].Computer Simulation,2009,26(10):327-329,343. |
| |
| Authors: | FENG Yun JING Xin-xing |
| |
| Affiliation: | Information and Communication College;Guilin University of Technology;Guilin Guangxi 541004;China |
| |
| Abstract: | MFCC(Me1-Frequency-Cepstral Coefficients)is based on the human ears' non-linear frequency characteristic and has a high recognition rate in practical speech recognition application.In order to further perfect speech recognition system and increase recognition rate effectively,an improved extraction algorithm for MFCC feature is presented.Firstly,the article compares compression results based on the method of using traditional linear MFCC feature with compression results based on combining evidence from resi... |
| |
| Keywords: | Speech recognition Mel frequency cepstral coefficient(MFCC) Residual phase Recognition rate |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
