| 基于相对相关序列MFCC特征的模型补偿技术 |
| |
| 引用本文： | 张军,韦岗. 基于相对相关序列MFCC特征的模型补偿技术[J]. 信号处理, 2003, 19(3): 284-286 |
| |
| 作者姓名： | 张军 韦岗 |
| |
| 作者单位： | 华南理工大学电子与信息学院,广州,510640 |
| |
| 基金项目： | 国家自然科学基金(60172048) |
| |
| 摘 要： | 提出了基于相对自相关序列(Relative Autocorrelation Sequences,RAS)MFCC特征的模型补偿技术,并给出其理论推导及分析。实验结果表明,本文提出的补偿技术可以在不同信噪比下提高RAS-MFCC的性能,使基于RAS-MFCC的识别系统在低信噪比时仍能保持较高的识别率,并且在低信噪比时其性能明显优于经过相同补偿处理的MFCC识别系统。
|
| 关 键 词： | 语音识别 相对自相关序列 模型补偿 |
| 修稿时间： | 2002-07-24 |
Model Compensation for MFCCs based on Relative Autocorrelation Sequences |
| |
| Zhang Jun Wei Gang. Model Compensation for MFCCs based on Relative Autocorrelation Sequences[J]. Signal Processing(China), 2003, 19(3): 284-286 |
| |
| Authors: | Zhang Jun Wei Gang |
| |
| Abstract: | This paper proposes a model compensation scheme for MFCCs based on relative autocorrelation sequences. Experimental results show that by the proposed compensation scheme the performance of RAS-MFCCs can be improved under a wide range of SNR for different kind of noises. With the same compensation scheme, RAS-MFCCs also outperforms MFCCs in low SNR levels. |
| |
| Keywords: | speech recognition relative autocorrelation sequences model compensation |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
