CN110111799B - AMR fixed codebook security steganography method based on pulse distribution model - Google Patents

AMR fixed codebook security steganography method based on pulse distribution model Download PDF

Info

Publication number
CN110111799B
CN110111799B CN201910347984.3A CN201910347984A CN110111799B CN 110111799 B CN110111799 B CN 110111799B CN 201910347984 A CN201910347984 A CN 201910347984A CN 110111799 B CN110111799 B CN 110111799B
Authority
CN
China
Prior art keywords
tag
information
mask
secret information
track
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201910347984.3A
Other languages
Chinese (zh)
Other versions
CN110111799A (en
Inventor
任延珍
杨瀚溢
吴红霞
王丽娜
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Wuhan University WHU
Original Assignee
Wuhan University WHU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Wuhan University WHU filed Critical Wuhan University WHU
Priority to CN201910347984.3A priority Critical patent/CN110111799B/en
Publication of CN110111799A publication Critical patent/CN110111799A/en
Application granted granted Critical
Publication of CN110111799B publication Critical patent/CN110111799B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/018—Audio watermarking, i.e. embedding inaudible data in the audio signal
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L9/00—Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols
    • H04L9/08—Key distribution or management, e.g. generation, sharing or updating, of cryptographic keys or passwords
    • H04L9/0861—Generation of secret information including derivation or calculation of cryptographic keys or passwords
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L9/00—Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols
    • H04L9/08—Key distribution or management, e.g. generation, sharing or updating, of cryptographic keys or passwords
    • H04L9/0861—Generation of secret information including derivation or calculation of cryptographic keys or passwords
    • H04L9/0869—Generation of secret information including derivation or calculation of cryptographic keys or passwords involving random numbers or seeds

Abstract

The invention discloses an AMR fixed codebook security steganography method based on a pulse distribution model. The corresponding embedding rule is designed according to the principle of minimizing the change of the pulse distribution characteristics, so that the probability that the pulse positions on the same track are the same is unchanged, and the distribution is random, so that the steganographic audio is close to the original audio in the pulse distribution. The invention has the advantages of good concealment and strong anti-steganalysis capability.

Description

AMR fixed codebook security steganography method based on pulse distribution model
Technical Field
The invention belongs to the technical field of multimedia information content security, relates to an AMR fixed codebook-oriented steganography method, and particularly relates to an AMR fixed codebook secure steganography method based on a pulse distribution model.
Technical Field
At present, with the rapid development of mobile internet technology and the popularization and application of various audio and voice conversation services in the mobile internet, various compressed audio files are shared and distributed on the internet, and a rich steganography space is provided for information hiding. Meanwhile, the traditional constant-rate speech coding format cannot meet the requirements of speech quality and speech transmission occupying less network bandwidth at the same time. AMR (Adaptive Multi-Rate code) is a speech compression coding standard specified by the 3GPP organization, and is currently widely used in GSM, TDMA, UMTS, and VOLTE, as well as in various mobile phone terminal systems, such as iPhone, Samsung, Nokia, and so on. In addition, some mainstream mobile terminal instant messaging software, such as QQ, WeChat, etc., also support the AMR coding format.
While AMR is widely used, it also provides a new rich carrier for covert communications. At present, steganographic schemes for AMR compressed audio are emerging, which embed secret information by modifying the encoder parameters during the encoding process. Based on algebraic code excited linear prediction principle, there are three feasible embedded domains in AMR compression coding, and the three embedded domains are respectively: a fixed codebook, linear prediction coefficients and a pitch period. The three embedding domains provide a great deal of redundancy for AMR steganography, and the existing steganography algorithm just utilizes the redundant information to embed secret information. The bit number used for pulse position parameter coding as the key part of the AMR fixed codebook accounts for more than 30% of each frame, and is the embedding domain with the largest hidden space in the AMR compressed domain.
According to the search principle of the AMR fixed codebook, the AMR fixed codebook has a large steganography space, and the existing steganography algorithm based on the AMR fixed codebook has two problems. The existing steganalysis algorithm utilizes the abnormal information to analyze, so that better detection performance is achieved. Secondly, the basic principle of AMR coding is not considered when the hidden writing is carried out, so the hidden writing causes the great reduction of the voice quality, and the voice quality is more obviously reduced under the condition of low code rate.
AMR coding is an ACELP-based hybrid speech compression coding technique in which an AMR coder adaptively selects a coding mode according to the channel characteristics of a sound source and the channel quality used to transmit a signal in order to improve the quality of the synthesized speech. Based on the principle that the weighted mean square error of the synthesized speech and the original speech is minimum, when the AMR is used for coding the original speech, the AMR searches the optimal codebook vectors for replacing residual signals from the adaptive codebook and the fixed codebook respectively, the codebook vectors are subjected to quantization coding and then are sent to a decoding end, and the decoding end can recover the synthesized speech close to the original speech according to the received parameters. Fig. 1 is a schematic diagram of the AMR coding principle. The AMR coding process mainly includes four parts, preprocessing, LP (Linear Predictive, LP) Linear prediction, fixed codebook search, and adaptive codebook search (also called pitch lag search). When coding, a frame is used as a unit for coding, a non-compressed linear PCM voice stream with 8kHz sampling rate and 16 bit quantization is input, in an AMR coding format, the voice time of each frame is 20ms and comprises 160 sampling points, and each frame is divided into 4 subframes by taking 5ms as a unit.
The invention content is as follows:
the invention provides an AMR fixed codebook secure steganography method based on a pulse distribution model, aiming at the problem that the existing AMR fixed codebook domain steganography algorithm is unsafe.
The technical scheme adopted by the invention is as follows: an AMR fixed codebook security steganography method based on a pulse distribution model is characterized by comprising the following steps:
step 1: determining an embedding rule;
Figure GDA0002834708890000021
wherein: t, 0. ltoreq. t.ltoreq.4 denotes a track number, itAnd it+5Representing two non-zero pulse positions in the track t,
Figure GDA0002834708890000022
meaning that the rounding is done down for x,
Figure GDA0002834708890000023
representing a bitwise XOR operation, mtRepresents 3-bit secret information to be embedded in the track t;
when embedding, firstly judging two non-zero pulse positions i in the track ttAnd it+5Whether or not they are the same, if it＝it+5Otherwise, the track is not steganographically written, otherwise, the track is steganographically written according to the embedded secret information mtEmbedding is carried out in two cases:
1) let the candidate pulse index satisfying equation (1) be k if k ≠ itAnd/5, directly taking k x 5+ t as a second non-zero pulse it+5The position of (a);
2) if k is equal to it5, the position of the second non-zero pulse is not changed;
step 2: secret information processing including secret information preprocessing, encoding and randomization;
the preprocessing is to encrypt and compress the secret information;
in the encoding, each 3-bit secret information is used as a preprocessing unit and is marked as msg, meanwhile, a 1-bit marking bit tag is used for marking each 3-bit secret information, whether the information is 0 is marked, if msg is 0, the corresponding marking bit tag is 0, and otherwise, tag is 1;
the randomization is to randomize tag by mask, and the process is as shown in formula 2:
Figure GDA0002834708890000031
wherein tagmaskMask is 3-bit mask for the masked marking information; let Ptag,Pmask,PtagmaskProbability distributions of 3-bit marking information, a mask and masked marking information respectively; in order not to change the probability of the same pulse occurring in the same track, tag is made simultaneouslymaskIs sufficiently random to satisfy the constraint shown in equation 3:
Figure GDA0002834708890000032
solving the equation shown in the formula 4 by constructing the probability distribution of the mask to ensure that tagmaskSatisfying the condition of formula 3; wherein i is more than or equal to 0, j is less than or equal to 7, PmaskProbability distribution to be solved; ptagThe values of (A) are as follows: when tag is 000, P tag1/512; when tag is 001, Ptag7/512; when tag is 010, Ptag7/512; when tag is 011, Ptag49/512; when tag is 100, Ptag7/512; when tag is 101, Ptag49/512; when tag is 110, Ptag49/512; when tag is 111, Ptag＝343/512；
Figure GDA0002834708890000033
Upon solving for PmaskThen, a random mask sequence can be generated, and the original tag information tag is randomized;
and step 3: the processed secret information is embedded into the pulse positions of the audio.
Compared with the prior art, the method and the device design the embedding process and the random check code by analyzing the distribution characteristics of the AMR voice fixed codebook so as to ensure that the distribution of the AMR voice after steganography is close to that of the original voice. Therefore, when the steganalysis algorithm is resisted, the voice generated by the method is difficult to detect, and the safety is greatly improved.
Drawings
FIG. 1 is a schematic diagram of AMR speech coding in the background art;
FIG. 2 is a flowchart of secret information encoding processing according to an embodiment of the present invention;
fig. 3 is a flowchart of generating a random mask sequence according to an embodiment of the present invention.
Detailed Description
In order to facilitate the understanding and implementation of the present invention for those of ordinary skill in the art, the present invention is further described in detail with reference to the accompanying drawings and examples, it is to be understood that the embodiments described herein are merely illustrative and explanatory of the present invention and are not restrictive thereof.
Referring to fig. 1, the AMR fixed codebook security steganography method based on the pulse distribution model provided by the present invention includes the following steps:
step 1: determining an embedding rule;
Figure GDA0002834708890000041
wherein: t, 0. ltoreq. t.ltoreq.4 denotes a track number, itAnd it+5Representing two non-zero pulse positions in the track t,
Figure GDA0002834708890000042
meaning that the rounding is done down for x,
Figure GDA0002834708890000043
representing a bitwise XOR operation, mtRepresents 3-bit secret information to be embedded in the track t;
when embedding, firstly judging two non-zero pulse positions i in the track ttAnd it+5Whether or not they are the same, if it＝it+5Otherwise, the track is not steganographically written, otherwise, the track is steganographically written according to the embedded secret information mtEmbedding is carried out in two cases:
1) recall that the candidate pulse index satisfying equation (1) isk, if k ≠ itAnd/5, directly taking k x 5+ t as a second non-zero pulse it+5The position of (a);
2) if k is equal to it5, the position of the second non-zero pulse is not changed;
step 2: and secret information processing is used for ensuring that the pulse distribution of the original audio is not damaged in the embedding process and improving the safety. Preprocessing, encoding and randomizing secret information;
and preprocessing, namely encrypting and compressing the secret information, so that the secret information is distributed more randomly and the algorithm is safer.
Encoding, according to the principle of bitwise XOR operation and the rule of embedding, if secret information msg is to be embeddedtTo ensure that the probability of the same pulse position occurring on the uniform track is constant, the second pulse position is not modified. In order to correctly extract the secret information in this case, it is necessary to perform encoding processing on the original secret information in advance. Taking each 3-bit secret information as a preprocessing unit, marking the preprocessed information as msg, marking each 3-bit secret information by using a 1-bit marking bit tag, and marking whether the information is 0, wherein if msg is 0, the corresponding marking bit tag is 0, otherwise, tag is 1; the encoding process is shown in fig. 2;
randomization, the encoded tag information tag will be embedded into the audio track in units of every 3 bits in order to ensure correct extraction of the secret information. However, the distribution of tag is not uniform every 3 bits, and the probability distribution is shown in table 1. Therefore, simply embedding will destroy the original pulse distribution of the audio, so it needs to randomize tag by mask, as shown in equation 2:
Figure GDA0002834708890000051
wherein tagmaskMask is 3-bit mask for the masked marking information; let Ptag,Pmask,PtagmaskProbability distributions of 3-bit marking information, a mask and masked marking information respectively; in order not to change in the same trackProbability of same pulse occurring, while making tagmaskIs sufficiently random to satisfy the constraint shown in equation 3:
Figure GDA0002834708890000052
solving the equation shown in the formula 4 by constructing the probability distribution of the mask to ensure that tagmaskSatisfying the condition of formula 3; wherein i is more than or equal to 0, j is less than or equal to 7, PmaskProbability distribution to be solved; ptagAs shown in table 1;
Figure GDA0002834708890000053
TABLE 1 probability distribution of three-bit tag information tag
tag 000 001 010 011 100 101 110 111
P tag 1/512 7/512 7/512 49/512 7/512 49/512 49/512 343/512
Upon solving for PmaskThen, a random mask sequence can be generated, and the original tag information tag is randomized;
the process of generating the mask sequence is shown in fig. 3; first according to PmaskAnd generating a mask sequence with the length of N, and then carrying out disorder processing on the mask sequence according to the key selected by the user.
And step 3: embedding the processed secret information into the pulse position of the audio;
the specific implementation of the step 3 comprises the following substeps:
step 3.1: dividing the processed secret information into original secret information and mark information;
step 3.2: when AMR voice is coded according to frames, fixed codebook search is carried out to obtain two nonzero pulse positions i in each track ttAnd it+5；
Step 3.3: embedding according to the type of the embedded information and a corresponding embedding rule;
if the embedded information is original secret information, judging whether the two pulse positions in the track t are equal or not; if it＝it+5Skipping the track to the next track t +1, and turning to the step 3.2; if it≠it+5Embedding according to the original secret information embedding rule;
if the embedded information is the mark information, the two pulse positions in the track t do not need to be judged to be equal, and the embedding is directly carried out according to the embedding rule;
step 3.4: and coding frame by frame, and embedding according to step 3.3 until finishing.
In this embodiment, during decoding, each 8 groups of information (each group includes 3 bits of information) are decoded as a group, and the process is as follows:
step 1: decoding the audio to obtain two non-zero pulse positions in each track;
step 2: extracting the secret information and the mark information according to the embedding rule of the secret information and the mark information;
and step 3: modifying the secret information based on the tag information;
and 4, step 4: and decrypting the corrected secret information to obtain the original secret information.
The invention analyzes the pulse distribution characteristic in cover audio frequency, combines the AMR fixed codebook coding principle, preprocesses the secret information during embedding, and divides the embedded information into original secret information and mark information. The corresponding embedding rule is designed according to the principle of minimizing the change of the pulse distribution characteristics, so that the probability that the pulse positions on the same track are the same is unchanged, and the distribution is random, so that the steganographic audio is close to the original audio in the pulse distribution. The algorithm has the advantages of good concealment and strong anti-steganalysis capability.
It should be understood that parts of the specification not set forth in detail are well within the prior art.
It should be understood that the above description of the preferred embodiments is given for clarity and not for any purpose of limitation, and that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims (3)

1. An AMR fixed codebook security steganography method based on a pulse distribution model is characterized by comprising the following steps:
step 1: determining an embedding rule;
Figure FDA0002843096700000011
wherein: t, 0. ltoreq. t.ltoreq.4 denotes a track number, itAnd it+5Representing two non-zero pulse positions in the track t,
Figure FDA0002843096700000012
meaning that the rounding is done down for x,
Figure FDA0002843096700000013
representing a bitwise XOR operation, mtRepresents 3-bit secret information to be embedded in the track t;
when embedding, firstly judging two non-zero pulse positions i in the track ttAnd it+5Whether or not they are the same, if it＝it+5Otherwise, the track is not steganographically written, otherwise, the track is steganographically written according to the embedded secret information mtEmbedding is carried out in two cases:
1) let the candidate pulse index satisfying equation (1) be k if k ≠ itAnd/5, directly taking k x 5+ t as a second non-zero pulse it+5The position of (a);
2) if k is equal to it5, the position of the second non-zero pulse is not changed;
step 2: secret information processing including secret information preprocessing, encoding and randomization;
the preprocessing is to encrypt and compress the secret information;
in the encoding, each 3-bit secret information is used as a preprocessing unit and is marked as msg, meanwhile, a 1-bit marking bit tag is used for marking each 3-bit secret information, whether the information is 0 is marked, if msg is 0, the corresponding marking bit tag is 0, and otherwise, tag is 1;
the randomization is to randomize tag by mask, and the process is as shown in formula 2:
Figure FDA0002843096700000014
wherein tagmaskMask is 3-bit mask for the masked marking information; let Ptag,Pmask,PtagmaskProbability distributions of 3-bit marking information, a mask and masked marking information respectively; in order not to change the probability of the same pulse occurring in the same track, tag is made simultaneouslymaskIs sufficiently random to satisfy the constraint shown in equation 3:
Figure FDA0002843096700000015
solving the equation shown in the formula 4 by constructing the probability distribution of the mask to ensure that tagmaskSatisfying the condition of formula 3; wherein i is more than or equal to 0, j is less than or equal to 7, PmaskProbability distribution to be solved; ptagThe values of (A) are as follows: when tag is 000, Ptag1/512; when tag is 001, Ptag7/512; when tag is 010, Ptag7/512; when tag is 011, Ptag49/512; when tag is 100, Ptag7/512; when tag is 101, Ptag49/512; when tag is 110, Ptag49/512; when tag is 111, Ptag＝343/512；
Figure FDA0002843096700000021
Upon solving for PmaskThen, a random mask sequence can be generated, and the original tag information tag is randomized;
and step 3: embedding the processed secret information into the pulse position of the audio;
the specific implementation of the step 3 comprises the following substeps:
step 3.1: dividing the processed secret information into original secret information and mark information;
step 3.2: for AMR languageWhen coding according to frames, fixed codebook search obtains two nonzero pulse positions i in each track ttAnd it+5；
Step 3.3: embedding according to the type of the embedded information and a corresponding embedding rule;
if the embedded information is original secret information, judging whether the two pulse positions in the track t are equal or not; if it＝it+5Skipping the track to the next track t +1, and turning to the step 3.2; if it≠it+5Embedding according to the original secret information embedding rule;
if the embedded information is the mark information, the two pulse positions in the track t do not need to be judged to be equal, and the embedding is directly carried out according to the embedding rule;
step 3.4: and coding frame by frame, and embedding according to step 3.3 until finishing.
2. The AMR fixed codebook security steganography method based on a pulse distribution model according to claim 1, wherein the step 2 of generating a random mask sequence is implemented by: first according to PmaskAnd generating a mask sequence with the length of N, and then carrying out disorder processing on the mask sequence according to the key selected by the user.
3. The AMR fixed codebook secure steganography method based on a pulse distribution model according to any one of claims 1-2, wherein: during decoding, each 8 groups of information are decoded, and each group comprises 3 bits of information; the decoding process comprises the following sub-steps:
step 4.1: decoding the audio to obtain two non-zero pulse positions in each track;
step 4.2: extracting the secret information and the mark information according to the embedding rule of the secret information and the mark information;
step 4.3: modifying the secret information based on the tag information;
step 4.4: and decrypting the corrected secret information to obtain the original secret information.
CN201910347984.3A 2019-04-28 2019-04-28 AMR fixed codebook security steganography method based on pulse distribution model Active CN110111799B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910347984.3A CN110111799B (en) 2019-04-28 2019-04-28 AMR fixed codebook security steganography method based on pulse distribution model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910347984.3A CN110111799B (en) 2019-04-28 2019-04-28 AMR fixed codebook security steganography method based on pulse distribution model

Publications (2)

Publication Number Publication Date
CN110111799A CN110111799A (en) 2019-08-09
CN110111799B true CN110111799B (en) 2021-02-19

Family

ID=67487166

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910347984.3A Active CN110111799B (en) 2019-04-28 2019-04-28 AMR fixed codebook security steganography method based on pulse distribution model

Country Status (1)

Country Link
CN (1) CN110111799B (en)

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7363218B2 (en) * 2002-10-25 2008-04-22 Dilithium Networks Pty. Ltd. Method and apparatus for fast CELP parameter mapping
US7231271B2 (en) * 2004-01-21 2007-06-12 The United States Of America As Represented By The Secretary Of The Air Force Steganographic method for covert audio communications
CN106101714B (en) * 2016-07-14 2019-06-04 中国科学院信息工程研究所 A kind of and tightly coupled H.264 video information hiding method of compression encoding process
CN107689226A (en) * 2017-08-29 2018-02-13 中国民航大学 A kind of low capacity Methods of Speech Information Hiding based on iLBC codings
CN107564536A (en) * 2017-09-06 2018-01-09 武汉大学 AMR pitch delay steganalysis methods based on difference Markov transition probability features in pitch delay subframe group group

Also Published As

Publication number Publication date
CN110111799A (en) 2019-08-09

Similar Documents

Publication Publication Date Title
JP6199334B2 (en) Equipment for encoding and detecting watermarked signals
Sun et al. High performance reversible data hiding for block truncation coding compressed images
Ren et al. AMR steganalysis based on the probability of same pulse position
Chou et al. On the duality between distributed source coding and data hiding
US8412519B2 (en) Steganography in digital signal encoders
Wu et al. An approach of steganography in G. 729 bitstream based on matrix coding and interleaving
EP2673769B1 (en) Devices, methods and computer-program product for adaptively encoding and decoding a watermarked signal
US20080140392A1 (en) Codec mode decoding method and apparatus for adaptive multi-rate system
Yin et al. Robust adaptive steganography based on dither modulation and modification with re-compression
CN107886962A (en) A kind of high security steganography method towards ip voice
Kheddar et al. High capacity speech steganography for the G723. 1 coder based on quantised line spectral pairs interpolation and CNN auto-encoding
CN103002406B (en) A kind of voice encryption method being applied to arrowband radio digital communication system
Ren et al. A secure AMR fixed codebook steganographic scheme based on pulse distribution model
Tian et al. Detecting bitrate modulation-based covert voice-over-IP communication
Deepikaa et al. VoIP steganography methods, a survey
CN110111799B (en) AMR fixed codebook security steganography method based on pulse distribution model
CN112714230A (en) Robust video steganography method and device based on audio side channel
Özkan et al. Data transmission via GSM voice channel for end to end security
Kazemi et al. Data hiding robust to mobile communication vocoders
CN107545899B (en) AMR steganography method based on unvoiced fundamental tone delay jitter characteristic
Kheddar et al. Fourier magnitude-based steganography for hiding 2.4 kbpsmelp secret speech
CN102918589B (en) The audio or video signal of decoding of the watermark derived from the audio or video signal of the watermark with low rate encoding and decoding is provided
US20090204393A1 (en) Systems and Methods For Adaptive Multi-Rate Protocol Enhancement
Li et al. Audio steganography algorithm based on genetic algorithm for MDCT coefficient adjustment for AAC
CN104867496A (en) MDCT quantization coefficient small value region-based advanced audio coding (AAC) audio steganography and extraction method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant