Research On AMR Steganography In FCB Domain

As the 3GPP organization designed voice compression standard,AMR is currently widely used in a variety of mobile phone terminal system.AMR provides a new carrier for secret communication.Based on the principle of Algebraic Code Excited Linear Prediction(ACELP),the FCB parameters are important part of the AMR coding parameters.FCB is generated by the non-exhaustive depth-first tree search,so there are many redundancy in FCB parameters,so it provides a new covert steganography space.The existing AMR FCB steganography sachems never consider the FCB search principle,so the embedding will cause the obvious difference between the statics characters of cover and stego,and the existing steganography algorithms cannot resist the existing steganalysis algorithm.Therefore,this paper studies the pulse distribution of AMR FCB to improve the hearing concealment and statistical security.The main works and innovations in this paper are as follows:1.Aiming at the problem that the existing AMR FCB steganography algorithms has low concealment,an adaptive steganography algorithm for AMR based on the optimal pulse and pulse correlation is proposed.Based on the search principle of AMR FCB and the correlation of non-zero pulse positions,the algorithm designs the cost function and the additive distortion function.Combined with STC the algorithm achieves adaptive steganography.In 12.2 kbit/s mode,the PESQ of stegogenerated by the proposed algorithm is 8%more than the existing steganography schemes.The average test error rate(TER)of the proposed algorithm is improved about 10%than the existing steganography algorithms.The experiment results show that the AFAS algorithm has better hearing concealment,and is more secure to hide from the steganalysis algorithms in literatures.2.Aiming at the problem of low statistical security of existing AMR FCB steganography algorithms,a security steganography algorithm(PDM SAFS)based on pulse distribution and an improved steganography algorithm based on random mask(RM_PDM SAFS)are proposed.In order to make the pulse distribution of stego closer to cover,the algorithm preprocesses the secret information to divide the original secret information into embedded information and flag information.Then PDM SAFS algorithm designs the corresponding steganography strategy respectively.Finaly,the PDM_SAFS algorithm is improved by using random mask to preprocess the flag information.The experimental results show that the error rate of stego of the PDM_SAFS algorithm and the RM_PDM_SAFS algorithm is about 25%with the relative embedding rate of 100%.And the TER of cover audios of RM_PDM_SAFS algorithm is 5%higher than that of PDM SAFS algorithm.So the PDM SAFS algorithm and RM_PDM_SAFS aogorithm have higher statistical security.In addition,the hidden capacity of the PDM SAFS algorithm and the RM_PDM_SAFS algorithm is relatively high too.In 12.2kbit/s mode,the maximum embedding capacity of PDM_SAFS algorithm is 2.25kbps.3.For AMR fixed codebook,by changing the non-zero pulse position to study the impact of each pulse on the audio quality and the hidden capacity of the FCB domain.For the two coding modes of 12.2kbit/s and 10.2kbit/s,the paper detects the hearing concealment and statistical security of stego generated by modifying the non-zero pulse position one by one.For the other coding modes,the paper analyzes the hearing concealment and the histogram,entropy and variance of pulse distribution of stego audios.The experimental results show that the last non-zero pulse in each track is the ideal embedding position for 12.2kbit/s and 10.2kbit/s modes.And in other coding modes,besides the first non-zero pulse,the other non-zero pulse positions are better embedding positions.In this paper,for fixed codebook,we propose an adaptive steganography algorithm(AFAS)and a security capacity steganography algorithm(PDM SAFS)and an improved steganography algorithm(RM_PDM_SAFS).In addition,the paper studies the hidden capacity of AMR FCB domain.The research results of the paper provide technical support for secure secret communication.