| 基于多窗频谱估计和平滑幅度谱包络的MFCC改进算法 |
| |
| 引用本文： | 张怡然,白静,王力. 基于多窗频谱估计和平滑幅度谱包络的MFCC改进算法[J]. 科学技术与工程, 2014, 14(19) |
| |
| 作者姓名： | 张怡然 白静 王力 |
| |
| 作者单位： | 太原理工大学 信息工程学院,太原理工大学 信息工程学院,太原理工大学 信息工程学院 |
| |
| 基金项目： | 国家自然科学基金资助项目(61072087)；山西省科技攻关项目(20120313013-6);山西省青年科技研究基金(2013021016-1) |
| |
| 摘 要： | 语音的特征提取是说话人识别系统中的关键问题。本文在传统的MFCC参数的基础上,提出一种改进的MFCC特征提取算法。该算法着眼于语音的前端处理,在预处理阶段,利用SWCE窗函数,对信号进行多窗频谱估计。并对得到的频谱进行平滑处理,得到信号的谱包络。然后对信号的谱包络进行计算,得到改进的MFCC参数。实验表明,在不同噪声环境下,与传统的MFCC算法相比,改进的算法识别率提高四个百分点以上。
|
| 关 键 词： | MFCC 多窗频谱估计 滑动平均滤波 谱包络 说话人识别 |
| 收稿时间： | 2014-01-21 |
| 修稿时间： | 2014-03-06 |
An Improved MFCC Algorithm Based on Multiple Window Spectrum Estimation and Smoothing Spectrum Envelope |
| |
| ZHANG Yiran,BAI Jing and WANG Li. An Improved MFCC Algorithm Based on Multiple Window Spectrum Estimation and Smoothing Spectrum Envelope[J]. Science Technology and Engineering, 2014, 14(19) |
| |
| Authors: | ZHANG Yiran BAI Jing WANG Li |
| |
| Affiliation: | College of Information Engineering,Taiyuan University of Technology,College of Information Engineering,Taiyuan University of Technology |
| |
| Abstract: | Feature extraction of speech is a key problem in speakerrecognition system. This article proposes a improved MFCC feature extraction algorithm which is based on the traditional MFCC parameters. TheSalgorithm focuses on the front-end processing. In the preprocessing stage, use the SWCESwindow function, to get theSmultiple windowSspectrumSestimation. And smooth the obtained spectrum, to get envelopeSsignal. ThenScalculate the spectrum envelopeSsignal,Sto getSthe improved MFCC parameters. Experiments show that,Sin the noise environment,Scompared with the traditional MFCC algorithm,Sthe recognition rate of the improvedSalgorithmSis improved more than four percent . |
| |
| Keywords: | MFCC multitapering moving average filter spectral envelope speaker recognition |
|
| 点击此处可从《科学技术与工程》浏览原始摘要信息 |
|
点击此处可从《科学技术与工程》下载全文 |
|
