Computer Engineering and Applications ›› 2009, Vol. 45 ›› Issue (33): 108-110.DOI: 10.3778/j.issn.1002-8331.2009.33.035

• 数据库、信号与信息处理 • Previous Articles     Next Articles

Improved speech recognition algorithm based on MFCC feature

YUAN Zheng-wu1,2,XIAO Wang-hui1   

  1. 1.Sino-Korea GIS Research Center,Chongqing University of Posts & Telecommunications,Chongqing 400065,China
    2.Civil Engineering Mobile Station for Post Doctors,Chongqing University,Chongqing 400045,China
  • Received:2009-06-15 Revised:2009-08-04 Online:2009-11-21 Published:2009-11-21
  • Contact: YUAN Zheng-wu

改进的混合MFCC语音识别算法研究

袁正午1,2,肖旺辉1   

  1. 1.重庆邮电大学 中韩合作GIS研究所,重庆 400065
    2.重庆大学 土木工程学院博士后流动站,重庆 400045
  • 通讯作者: 袁正午

Abstract: Aiming at low calculation accuracy of MFCC feature parameters for speech recognition in high-frequency signals,this paper adopts the combination algorithm of IMFCC,MIDMFCC,MFCC,which uses a mixture of filter banks to enhance high-frequency calculation accuracy in the calculation accuracy.The experiments result shows that the recognition rate of improved algorithm increases compared to the classical algorithm in the same environment.

Key words: Mel-Frequency Cepstral Coefficients(MFCC), speech recognition, feature extraction

摘要: 针对MFCC特征参数在语音识别中对中高频信号的识别精度不高的特点,提出采用IMFCC,MIDMFCC,MFCC相结合的改进算法,使用混合滤波器组,提高在语音中高频区域中的识别精度。实验结果表明,改进之后的算法与经典算法比较,在相同环境下对语音信息的识别率都有一定程度的提高。

关键词: Mel频率倒谱系数(MFCC), 语音识别, 特征提取

CLC Number: