摘要
特征提取是说话人识别系统中的一项关键技术,讨论了在说话人识别中语音特征的提取,详细介绍了求取MEL倒谱系数MFCC和线性预测倒谱系数LPCC的具体步骤和算法,并对MFCC和LPCC语音特征参数做出了理论分析和实验数据比较,通过实验论证了对于低频语音,采用MFCC参数的说话人识别在屏蔽噪音和抗噪声能力都优于采用LPCC参数的说话人识别.
The feature extraction is a key technology in the speaker identification system, The feature parameter extraction is discussed in speaker identification system, also the principle of the feature extraction by feature analysis is proposed. The comparison is made between them. Through analysis of the MFCC and LPCC feature parameter, it is proved that for low frequency, MF- CC is better than LPCC in noise shielding and noise immunity.
出处
《河南工程学院学报（自然科学版）》
2010年第2期51-55,共5页
Journal of Henan University of Engineering：Natural Science Edition
基金
西华大学"西华杯"学生课外学术科技项目
西华大学校青年基金(Q0920211)
关键词
说话人识别
特征提取
MFCC特征参数
LPCC特征参数
speaker identification system
feature extraction
MFCC feature parameter
LPCC feature parameter
