针对LSB隐写的语音隐写分析方法  

Speech Steganalysis for LSB Steganography

作　　者：杨勇[1,2] 邓伟 李晶 朱作付 贾伟伟[1] 王晨[1] 张岩 YANG Yong;DENG Wei;LI Jing;ZHU Zuofu;JIA Weiwei;WANG Chen;ZHANG Yan(School of Information Engineering,Xuzhou College of Industrial Technology,Xuzhou Jiangsu 221140,China;Key Laboratory of Underwater Acoustic Signal Processing of Ministry of Education,Southeast University,Nanjing Jiangsu 210096,China;School of Information Engineering,Nanjing Audit Unversity,Nanjing Jiangsu 211815,China)

机构地区：[1]徐州工业职业技术学院信息工程学院,江苏徐州221140 [2]东南大学水声信号处理教育部重点实验室,江苏南京210096 [3]南京审计大学信息工程学院,江苏南京211815

出　　处：《电子器件》2023年第3期790-794,共5页Chinese Journal of Electron Devices

基　　金：国家自然科学基金项目(52005267)。

摘　　要：对常用的最低有效位(LSB)隐写技术进行了隐写分析。在目前的网络环境中,检测低嵌入率语音隐写信号仍然是一个非常有挑战性的课题。近年来,神经网络模型在许多课题中都取得了显著的性能。神经网络的主流架构包括卷积神经网络(CNNs)和循环神经网络(RNNs),这两种网络采用了不同的方式来理解各种信号。提出了一种合适的方法来结合这两种架构的优点,然后构造了一个新的模型,即CNN-LSTM网络来检测基于LSB的隐写方法。在所提出的模型中,使用双向长短时记忆循环神经网络(BiLSTM)来捕获信号的长时间上下文信息,然后使用CNN捕获局部特征和全局特征。实验结果表明,该模型相较对比方法对于基于LSB的隐写分析达到了更好的效果。The least significant bit(LSB)steganography is analyzed. In the current network environment, the detection of low embedding rate speech steganographic signal is still a very challenging topic. In recent years, neural network model has achieved remarkable performance in many subjects. The mainstream architectures of neural networks include convolution neural networks(CNNs)and recurrent neural networks(RNNs),which use different ways to understand various signals. An appropriate method is proposed to combine the advantages of the two architectures, and then a new model i.e.,CNN-LSTM network is constructed to detect LSB based steganography. In the proposed model, BiLSTM is used to capture the long-term context information of the signal, and CNN is used to capture the local and global features. The experimental results show that the model achieves better effect than the contrast method for LSB based steganalysis.

关 键 词：语音隐写分析 深度神经网络 最低有效位隐写 

分 类 号：TN912.3[电子电信—通信与信息系统] TP183[电子电信—信息与通信工程]