一种DCT域的音频信息隐藏方法  

AN AUDIO INFORMATION HIDING METHOD IN DCT DOMAIN

作　　者：王俊杰[1] 张晓明[2] 梅东霞[1] 

机构地区：[1]北京化工大学信息科学与技术学院,北京100029 [2]北京石油化工学院信息工程学院,北京102617

出　　处：《计算机应用与软件》2006年第6期33-35,共3页Computer Applications and Software

基　　金：北京市教委科技发展计划面上项目(KM200510017006)

摘　　要：基于信息隐藏技术,提出并实现了一种语音保密通信的新方法。首先将保密语音进行混合激励线性预测(MELP)编码,形成隐藏信息;然后对公开语音信号作分段DCT变换,隐藏信息通过量化处理嵌入到DCT域的中频点;解码器在同样的嵌入点盲提取出隐藏信息,最后经MELP解码恢复保密语音。实验结果表明该算法在隐藏信息后,信道中传输的公开语音具有透明性,并且在受到压缩、滤波等攻击时具有较高的鲁棒性。Based on the method of information hiding,a novel algorithm for security speech communication is designed and implemented in this paper. First,Mix-Excitation Linear Prediction（MELP） coding is used for security speech, Then the public speech signal is converted into segments and discrete cosine transform is performed,and the hiding bit is embedded in the intermediate frequency point of Discrete Consine Transform（DCT） by quantizing process. Detector extracts the embedded information without the original digital audio signal, and recovers security speech by MELP decoder. Experiment results show the speech which is embedded with information has transparent feature, and algorithm is strongly robust to many attacks such as compression, filter and soon.

关 键 词：混合激励线性预测 语音保密通信 信息隐藏 离散余弦变换 盲提取 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TN916.33[自动化与计算机技术—计算机科学与技术]