降采样固定时延估算法在声回波对消系统中的应用

山东大学学报(工学版) ›› 2011, Vol. 41 ›› Issue (3): 42-45.

降采样固定时延估算法在声回波对消系统中的应用

王心一1,杜光2*

1.国家专利局通信发明审查部, 北京 100088; 2. 山东大学软件学院, 山东济南 250101

收稿日期:2011-03-28 出版日期:2011-06-16 发布日期:2011-03-28
通讯作者: 杜光(1982- ),男,山东章丘人,硕士研究生,主要研究方向为公共安全.Email:satand@mail.sdu.edu.cn E-mail:satand@mail.sdu.edu.cn
作者简介:王心一(1980- ),女,山东济南人,博士,主要研究方向为语音信号处理、回波消除. Email:wxybear@126.com

Sub-sampling in flat delay estimation with application to an acoustic
echo cancellation system

WANG Xin-yi1, DU Guang2*

1. Communication Examination Department, State Intellectual Property Office of People′s Republic of China,
Beijing 100088, China; 2. Software College,Shandong University, Jinan 250101, China

Received:2011-03-28 Online:2011-06-16 Published:2011-03-28

摘要/Abstract

摘要：

提出一种基于降采样的固定时延估计算法,该算法在进行时延估计之前先对语音信号进行白噪声化和降采样,有效抑制互相关函数中的局部极大值旁瓣,克服了传统归一化互相关函数时延估计算法应用于声学回波对消系统中时估计误差大、鲁棒性差的问题；并以基于降采样的固定时延估算法为基础,设计实现声学回波抵消系统。计算机仿真结果证明：该算法运算量低、收敛速度快、性能稳定。

关键词: 语音信号处理, 回波对消, 固定时延估计, 降采样滤波器, 归一化互相关

Abstract:

A sub-sampling flat delay estimation algorithm was proposed. The proposed algorithm performed speech signal whitening and desampling before the delay estimation. This arithmetic could effectively suppress the side lobe of the generalized crosscorrelation function and could overcome the robust problem in the conventional time delay estimation algorithm based on the generalized cross-correlation function. Finally, an acoustic echo cancellation system was designed with the proposed flat delay estimation module. Computer simulation showed that the proposed algorithm performed better than the conventional one.

Key words: speech signal processing, echo cancellation, flat delay estimation, desampling filter, generalized cross-correlation

王心一1,杜光2*. 降采样固定时延估算法在声回波对消系统中的应用[J]. 山东大学学报(工学版), 2011, 41(3): 42-45.

WANG Xin-yi1, DU Guang2*. Sub-sampling in flat delay estimation with application to an acoustic
echo cancellation system[J]. JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE), 2011, 41(3): 42-45.

参考文献

多维度评价

Viewed

Full text

Abstract

Cited

Shared

Discussed