藏语音存在语料库缺少和地区方言较多等问题,因此关于藏语音的识别技术相对缺乏。基于此,提出一种使用卷积神经网络(Convolutional Neural Network,CNN)、长短期记忆(Long Short Term Memory,LSTM)神经网路和动态神经网络(Dynamic Neura...藏语音存在语料库缺少和地区方言较多等问题,因此关于藏语音的识别技术相对缺乏。基于此,提出一种使用卷积神经网络(Convolutional Neural Network,CNN)、长短期记忆(Long Short Term Memory,LSTM)神经网路和动态神经网络(Dynamic Neural Network,DNN)的基于Python平台上TensorFlow框架的深度混合网络模型。首先,录制来自拉萨市、安多县和昌都市3个地区的藏语音数据制作语音数据集,并通过改进模型深度、结构、参数和算法来提升藏语音识别的准确率;其次,使用多层卷积残差网络和改进的LSTM神经网络解决模型训练过程中的梯度爆炸问题;最后,使用反向传播算法提高模型训练的准确度。仿真实验表明,该模型虽然在不同地区的藏语音数据识别准确率上存在差异,但是在整体的识别准确率和模型的收敛性上具有不错的效果。展开更多
The proposed secure communication approach adopts the proposed algorithm of Analysis-By- Synthesis （ABS） speech information hiding to establish a Secret Speech Subliminai Channel （SSSC） for speech secure communica...The proposed secure communication approach adopts the proposed algorithm of Analysis-By- Synthesis （ABS） speech information hiding to establish a Secret Speech Subliminai Channel （SSSC） for speech secure communication over PSTN （Public Switched Telephone Network）, and employs the algorithm of ABS speech information extracting to recovery the secret information, This approach is more reliable, covert and securable than traditional and chaotic secure communication.展开更多
In this paper, a Covert Speech Telephone (CST) is designed and implemented based on the information hiding technique, which works on the internet. To solve the large embedding capacity problem for real-time informatio...In this paper, a Covert Speech Telephone (CST) is designed and implemented based on the information hiding technique, which works on the internet. To solve the large embedding capacity problem for real-time information hiding, a steganographic system combined with a watermarking scheme is proposed, which skillfully transfers the secret speech into watermarking information. The basic idea is to use the speech recognition to significantly reduce the size of information that has to be transmitted in a hidden way. Furthermore, an improved DFT watermarking scheme is proposed which adaptively chooses the embedding locations and applies the multi-ary modulation technique. Based on the GUI (Graphical User Interface) software, the CST operates on both ordinary and secure mode. It is a completely digital system with high speech quality. Objective and subjective tests show that the CST is robust against normal signal processing attacks and steganalysis. The proposed scheme can be used in terms of military applications.展开更多
This paper presents a new spatial steganography model for covert communications over voice over internet protocol(Vo IP), providing a solution to the issue of increasing the capacity of covert Vo IP channels without c...This paper presents a new spatial steganography model for covert communications over voice over internet protocol(Vo IP), providing a solution to the issue of increasing the capacity of covert Vo IP channels without compromising the imperceptibility of the channels. Drawing from Orthogonal Modulation Theory in communications, the model introduced two concepts, orthogonal data hiding features and data hiding vectors, to covert Vo IP communications. By taking into account the variation characteristics of Vo IP audio streams in the time domain, a hiding vector negotiation mechanism was suggested to achieve dynamic self-adaptive steganography in media streams. Experimental results on Vo IP steganography show that the proposed steganographic method effectively depicted the spatial and temporal characteristics of Vo IP audio streams, and enhanced robustness against detection of steganalysis tools, thereby improving the security of covert Vo IP communications.展开更多
文摘藏语音存在语料库缺少和地区方言较多等问题,因此关于藏语音的识别技术相对缺乏。基于此,提出一种使用卷积神经网络(Convolutional Neural Network,CNN)、长短期记忆(Long Short Term Memory,LSTM)神经网路和动态神经网络(Dynamic Neural Network,DNN)的基于Python平台上TensorFlow框架的深度混合网络模型。首先,录制来自拉萨市、安多县和昌都市3个地区的藏语音数据制作语音数据集,并通过改进模型深度、结构、参数和算法来提升藏语音识别的准确率;其次,使用多层卷积残差网络和改进的LSTM神经网络解决模型训练过程中的梯度爆炸问题;最后,使用反向传播算法提高模型训练的准确度。仿真实验表明,该模型虽然在不同地区的藏语音数据识别准确率上存在差异,但是在整体的识别准确率和模型的收敛性上具有不错的效果。
基金Supported by the National High Technology Research and Development Program of China (863 Program) (No.2003AA142080, 2004AA775060)the National Natural Sicence Foundation of China (No.60203004)+1 种基金with additional support from the China Post-doctorial Research Foundation (2005-03)the Foundation of Tianjin Key Lab for Advanced Signal Processing(2005).
文摘The proposed secure communication approach adopts the proposed algorithm of Analysis-By- Synthesis （ABS） speech information hiding to establish a Secret Speech Subliminai Channel （SSSC） for speech secure communication over PSTN （Public Switched Telephone Network）, and employs the algorithm of ABS speech information extracting to recovery the secret information, This approach is more reliable, covert and securable than traditional and chaotic secure communication.
基金the Natural Science Foundation of Jiangsu Province (No.BK2004150)the National 863 Key Project (No.2006AA010102).
文摘In this paper, a Covert Speech Telephone (CST) is designed and implemented based on the information hiding technique, which works on the internet. To solve the large embedding capacity problem for real-time information hiding, a steganographic system combined with a watermarking scheme is proposed, which skillfully transfers the secret speech into watermarking information. The basic idea is to use the speech recognition to significantly reduce the size of information that has to be transmitted in a hidden way. Furthermore, an improved DFT watermarking scheme is proposed which adaptively chooses the embedding locations and applies the multi-ary modulation technique. Based on the GUI (Graphical User Interface) software, the CST operates on both ordinary and secure mode. It is a completely digital system with high speech quality. Objective and subjective tests show that the CST is robust against normal signal processing attacks and steganalysis. The proposed scheme can be used in terms of military applications.
基金supported by the National Natural Science Foundation of China(Grant Nos.61271392U1405254&61272469)the British Government(Grant No.ktp008263)
文摘This paper presents a new spatial steganography model for covert communications over voice over internet protocol(Vo IP), providing a solution to the issue of increasing the capacity of covert Vo IP channels without compromising the imperceptibility of the channels. Drawing from Orthogonal Modulation Theory in communications, the model introduced two concepts, orthogonal data hiding features and data hiding vectors, to covert Vo IP communications. By taking into account the variation characteristics of Vo IP audio streams in the time domain, a hiding vector negotiation mechanism was suggested to achieve dynamic self-adaptive steganography in media streams. Experimental results on Vo IP steganography show that the proposed steganographic method effectively depicted the spatial and temporal characteristics of Vo IP audio streams, and enhanced robustness against detection of steganalysis tools, thereby improving the security of covert Vo IP communications.
