高效的线性预测语音编码信息隐藏方法  

Efficient information hiding method for linear predictive coding based speech coder

作　　者：刘鹏[1,2] 李松斌[1,2] 戴琼兴[1] 邓浩江[2] 

机构地区：[1]中国科学院声学研究所南海研究站,海南海口570105 [2]中国科学院声学研究所国家网络新媒体工程技术研究中心,北京100190

出　　处：《计算机工程与设计》2014年第4期1172-1177,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61303249);国家863高技术研究发展计划基金项目(2011AA01A102);中科院战略性先导科技专项基金项目(XDA06010302)

摘　　要：为在线性预测语音编码器中进行信息隐藏,将语音码流中由线性预测系数矢量量化过程产生的所有可嵌入位置映射为一个信息隐藏空间,在空间中利用矩阵编码方法找出需要修改的位置并使用量化索引调制进行修改。引入了混沌理论以及可调安全等级的概念对嵌入位置及嵌入模版进行选择,进一步提高隐写的隐蔽性。实验结果表明,与现有方法相比,该方法在同等嵌入容量下具有失真小、隐蔽性高等优点。To hide information in linear predictive coding based speech coder, all the embeddable positions in speech stream pro- duced by vector quantization process of linear prediction coefficients are mapped into a steganography space. Then embedding is made by picking out the positions which needs to be modified for embedding in the steganography space by matrix coding and modifying them with quantization index modulation. Meanwhile, chaos theory and conception of security level are introduced for the selection of embedding positions and modes. Experimental results show that, compared with the existing method under the same embedding capacity, this method has better performance with less speech distortion and is more difficult to be detected.

关 键 词：信息隐藏 线性预测编码 矩阵编码 量化索引调制 混沌理论 

分 类 号：TP309.02[自动化与计算机技术—计算机系统结构]