CN101847409A - Voice integrity protection method based on digital fingerprint - Google Patents
Voice integrity protection method based on digital fingerprint Download PDFInfo
- Publication number
- CN101847409A CN101847409A CN201010132256A CN201010132256A CN101847409A CN 101847409 A CN101847409 A CN 101847409A CN 201010132256 A CN201010132256 A CN 201010132256A CN 201010132256 A CN201010132256 A CN 201010132256A CN 101847409 A CN101847409 A CN 101847409A
- Authority
- CN
- China
- Prior art keywords
- watermark
- data
- signal
- voice signal
- crc
- 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.)
- Granted
Links
Images
Landscapes
- Editing Of Facsimile Originals (AREA)
- Signal Processing For Digital Recording And Reproducing (AREA)
Abstract
The present invention discloses a voice integrity protection method based on digital fingerprint, which develops a relatively perfect method for the realization of audio integrity protection. The method first provides a combined watermarked data coding/decoding method based on trellis coding, BCH coding and CRC coding to improve the error code correcting performance of a system, then a watermark extraction synchronizing method using an energy synchronization head signal as a positioning standard and re-synchronization as complementary means to realize an accurate positioning on watermark extraction operation, and finally a voice integrity protection method taking watermark information as a primary determination criterion and the abstract of the watermark information as an auxiliary determination criterion to realize the core technology of voice integrity validation; according to the method, the user can transparently embed voice integrity protection information into a voice signal in real time during normal communication.
Description
Technical field
The present invention relates to multi-aspect information treatment technologies such as digital watermarking, voice communication, digital coding.In brief, the present invention proposes a kind of novel voice digital watermark technology implementation method, realized in voice communication course, be embedded in incompressible digital fingerprint data to voice signal in real time, and then provide integrity protection this application demand for voice signal.
Background technology
Recent two decades comes Personal Communications Services flourish, and landline telephone entered huge numbers of families and created communication environment easily for people before this, and mobile communication business has also been stepped into daily life subsequently, and interpersonal distance has further furthered.Under this environment, very big change has also taken place in the information acquisition mode of news media, and a large amount of telephone interviews continue to bring out; And, along with the development of digital audio processing technology with popularize, increasing layman just can edit the original audio material easily by general audio edited software; Though this working method improved widely the medium workers work efficiency, promoted the flourish of China's media business, also create the pseudo event condition of providing convenience for some illegal reporters.For properly settling this problem, the scientific worker must propose new technology solution route, and the digital watermark technology of rising in recent years provides new approaches for people.For example; when the reporter carries out telephone interview; if accessed people does not wish that some crucial language are updated by the reporter; the user can embed stronger integrity protection information of some robustnesss and copyright ownership information in real time at crucial speech position so, utilizes technological means to come the source and the integrality thereof of validating speech data so that occur after the dispute.
Find that through investigation there are following defective in traditional Audio Watermarking Algorithm and realization technology thereof.
The weak point of conventional audio watermarking algorithm:
Traditional Audio Watermarking Algorithm is based on the special audio data layout more, the specific compression algorithm is formulated, and their scope of application is comparatively narrow and small;
Mostly traditional Audio Watermarking Algorithm is to design at non real-time digital carriers such as various audio files, is not suitable for real-time communication system;
Though there are some traditional Audio Watermarking Algorithm robustnesss fine, number average can not be resisted the high strength voice scrambling compressed encoding attack in the voice communication system mostly.
The weak point of conventional audio watermarking algorithm implementation:
Mostly traditional Audio Watermarking Algorithm is to realize that its implementation platform does not possess portability on PC;
Though the part Audio Watermarking Algorithm can realize (for example PDA) on embedded intelligent terminal, because arithmetic speed is crossed the demand that can't satisfy real-time processing slowly;
The conventional audio watermarking algorithm can't be realized in a series of unintelligent terminal equipment such as landline telephone.
Current, the development trend of digital watermark technology is comparatively swift and violent, and the research that is introduced as Audio Watermarking Algorithm of human auditory system model provides the scientific basis on the human aesthesiology, and all kinds of watermarking algorithms also emerge in an endless stream.This provides broad theoretical foundation for the digital watermark technology application scheme that we study at voice communication system.The present invention is based on digital watermark technology, go through a large amount of research experiments after, a kind of voice integrity protection method based on digital finger-print has been proposed.
From implementing the required hardware supported platform of this method, more and more high-performance, low price, low-power consumption, embedded hardware equipment and chip are extensively popularized, and a kind of low cost, high-performance, the real-time treating apparatus of embedded digital signal provide powerful technical support for we develop for this.Integrated survey of the present invention existing digital audio processing scheme, through multi-argument and practice, on DSP embedded hardware platform, realized the exploitation and the application of digital watermark.
Current situation from the digital watermarking product, because digital watermark technology is not full-fledged as yet, most researchs only rest on the theoretical aspect, so the engineering practice scheme of moulding is actually rare, and the present invention proposes a kind of comparatively ripe digital watermark technology engineering application process.Through practical proof, the technology implementation method that the present invention proposes can satisfy the engineering application demand, the theoretical principle of digital watermark technology can be converted into application in practice, can be widely used in the various types of voice communication system.
Summary of the invention
The present invention proposes a kind of voice integrity protection method based on digital finger-print; this method has proposed the comparatively perfect digital watermark implementation of a cover around a kind of strong robustness watermarking algorithm; comprising watermarking algorithm implementation method, original watermark data-encoding scheme, voice signal completeness protection method, this method can be divided into watermark embedding method and watermark extracting method two parts content on the macroscopic view.
The digital audio frequency watermark technology implementation method that the present invention proposes can be described as briefly:
Communicating pair at first utilizes original communication system to set up normal voice communication;
Afterwards,, and utilize set watermarking algorithm that the voice signal that built-in end collected is handled, make it to carry watermark information by the real-time processing of watermark built-in end initiation to voice signal;
The voice signal that the watermark built-in end will be handled sends by original communication system;
After the watermark extracting termination is received the voice signal that carries watermark information, can utilize set watermark extracting algorithm to handle in real time and extract the voice signal that end is collected, and then obtain corresponding watermark information, also the speech data that receives can be preserved in order to a day aftertreatment;
But both call sides normal talking during this period.
The selected audio frequency watermark of this method embeds algorithm has been realized watermark information by the mode of in time domain speech data being divided into groups and adjust the signal energy ratio in each data set embedding, this watermarking algorithm can be resisted a RPE_LTP lossy compression method, twice AD conversion and interchannel noise and attack, have very strong robustness and transparent preferably (this method supports the user according to actual conditions, selects other strong robustness watermarking algorithm for use).Extract synchronization aspects at watermark data, this method has adopted the same footwork of energy head, and promptly the watermark extracting end is monitored the variation of voice signal in real time, and after the sign mutation pattern satisfied judgment condition, system started the method for watermark extraction process automatically.In addition, in the watermark information leaching process, extract point for guaranteeing the accurate locating information of system, this method has adopted heavy synchronization processing simultaneously, promptly in leaching process, system can readjust the method for watermark data extracting position according to the basic location information that the energy synchronous head signal is provided, and wherein adjusts step-length and setting range and can be made amendment according to practical application request by the user.Aspect data processing, in order to reduce the bit error rate of watermark information, improve the travelling speed of error correction algorithm, this method has adopted the combined encoding method that original watermark information is carried out Error Correction of Coding, thereby has improved the error code correction ability of system greatly.Aspect the integrity protection of voice signal, this method has at first increased the redundance of finger print information by the mode that embeds digital finger-print information repeatedly, has improved the anti-shearing attacking ability of holistic approach; Utilize the eap-message digest of one-way hash function calculating watermark data subsequently again, and come the integrality of aided verification watermark data, and then reflect the integrality of voice bearer with this summary info.
For proving absolutely that the technology that this method proposes realizes content, employed algorithm and variable in the method are done following symbolic representation agreement:
CRC (x) expression CRC cyclic redundancy check (CRC) function;
BCH (x) expression Bose-Chaudhuri-Hocquenghem Code function;
DBCH (x) expression BCH decoding and error function;
I (x) expression interweaving encoding function;
DI (x) expression deinterleaving coding function;
S
1(x) an expression watermark synchronization signal processing algorithm;
S
2(x) the heavy synchronous processing algorithm in the expression watermark extraction process;
Em (x, y) the expression digital watermarking embeds algorithm;
Re (x) expression digital watermarking extraction algorithm;
H (x) represents one-way hash function;
Data represents the original watermark data, the watermark data of the encoded algorithm process of Data ' expression, Data, and " watermark data that handle without decoding algorithm the back is extracted in expression;
Signal represents primary speech signal, the voice signal that Signal ' expression is handled through watermarking algorithm, Signal, and " expression Signal ' channel transmits the voice signal that the back is obtained by the watermark extracting end.
Token represents witness marker, and this sign is made up of two parts content---protected speech data section is encoded, the interior little group coding of watermark data of section, is used for indicating the segment number and the interior little group # of section of each group's watermark data.
(1) original watermark data-encoding scheme
The design concept of original watermark data-encoding scheme as shown in Figure 2, the method can be described as:
(1) watermark built-in end
Original watermark data Data will at first obtain CRC (Data)=Data+CRC by the CRC encoding process, and wherein CRC represents the check code of Data;
Afterwards, processing obtains BCH[CRC (Data) through Bose-Chaudhuri-Hocquenghem Code];
Finally, obtain Data '=I{BCH[CRC (Data) through interleaving treatment] }.
(2) watermark extracting end
The watermark information Data that extracts ", will at first obtain DI (Data ") by the deinterleaving encoding process;
Afterwards, through the BCH correction process obtain DBCH[DI (Data ")];
Finally, through the CRC cyclic redundancy check (CRC) handle obtain CRC{DBCH[DI (Data ")] };
If CRC check is correct, then can by CRC{DBCH[DI (Data ")] } restore Data;
If the CRC check mistake represents that then the bit error rate is too high, can't correctly recover original watermark data Data by above-mentioned Error Correction of Coding.
In the original watermark data-encoding scheme, to introduce the CRC cyclic redundancy check (CRC) and be in order to check the correctness of watermark data, and the BCH Error Correction of Coding to be in order correcting owing to the randomness watermark data mistake that wave form distortion causes takes place in transmission course the carrier voice signal; Because the randomness burst error mostly is continuous mistake in the channel, and this error pattern is unfavorable for BCH Error Correction of Coding performance useful effect, therefore in order to disperse continuous mistake, this method has been introduced the interweaving encoding processing, and wherein interleave depth can be provided with according to practical situations voluntarily by the user.
(2) watermarking algorithm implementation method
The implementation procedure of the selected watermarking algorithm of this method is an example with the telescopiny of any group watermark data bag as shown in Figure 3, and the method can be described as:
Both call sides is set up after the normal voice communication, at first by the watermark built-in end voice signal is handled, and obtains synchronous head signal S
1(Signal), this signal is an energy jump formula signal;
The watermark built-in end is finished synchronous head signal S
1(Signal) after the embedding operation, start watermark immediately and embed processing, utilize set watermarking algorithm to be embedded in the primary speech signal Signal watermark data Data ' to be embedded, thereby obtain carrying the voice signal Signal '=Em (Signal of watermark information, Data '), this process is real-time processing procedure;
The watermark extracting end is monitored the variation of voice signal in real time, as detect signal sudden change has taken place, and passed through the Signal Matching processing, satisfied judgement requirement (the watermark extracting end is preserved identical synchronous head signal sample), then the watermark extracting end will start watermark extraction process immediately, utilize set watermark extracting algorithm process signal Signal ", obtain Data "=Re (Signal ");
The watermark extracting end extracts Data " after, utilize data processing method shown in Figure 2 that Data " is carried out decoding processing; If CRC{DBCH[DI (Data ")] } verification is correct, then can correctly recover original watermark data Data;
If CRC{DBCH[DI (Data ")] } check errors, then the watermark extracting end need weigh the voice signal that collects synchronously and adjust, and can obtain S by moving the extraction anchor point minutely
2(Signal ");
Afterwards, the watermark extracting end will utilize set watermark extracting algorithm to obtain Data "=Re[S once more
2(Signal ")], and attempt recovering original watermark data Data;
If after making repeated attempts, the watermark extracting end fails to recover original watermark data Data, then watermark extraction process failure; If after making repeated attempts, the watermark extracting end recovers original watermark data Data, then writes down correct extracting position, provides reference point for extract the location next time.
The energy jump formula watermark synchronization head design that this method adopted can improve the success ratio of simultaneous extraction greatly, the design has taken into full account the influence of high compression ratio voice channel coding to the voice signal waveform, utilizes high energy signal to be subjected to the little characteristic of speech coding influence successfully to realize the synchronized positioning of watermark extracting; Find through practice, the synchronized positioning point can drift about along with the growth in transmission time in the watermark extraction process, and then have influence on the correct extraction and the recovery of watermark data, so this method has adopted heavy synchronous processing measure, promptly in watermark extraction process, constantly the synchronized positioning point is carried out small adjustment, revise the anchor point skew in time with this, realized the precise synchronization of watermark extracting.
(3) voice signal completeness protection method
For the to adequately protect integrality of voice signal of the digital watermark that this method is proposed; relevant watermark information should have stronger robustness (the particularly ability of shearing resistance cut attack); to guarantee that watermark extracting equipment can guarantee that the user can judge clearly by the watermark information that extracts whether corresponding voice signal is complete from being recovered the part watermark data the voice signal of montage.For satisfying above-mentioned application demand, the present invention proposes perfect voice signal completeness protection method, Fig. 4 is that example has provided result with any one section protected voice signal, can be described as in conjunction with legend this method:
After watermark built-in end, user's initiation are handled the protection of voice signal, system will embed one section watermark synchronization signal S in the reference position of protected voice signal
1(Signal)---this signal duration is very short, and is less to the auditory effect influence of normal voice signal;
After synchronous head signal embeds and finishes, system will pass through witness marker Token that digital coding handles, digital audio finger print information, " timestamp " information immediately and be embedded among the protected voice signal as the original watermark data, obtain carrying the voice signal Signal '=Em (Signal of digital finger-print information, Data '), thus the embedding of finishing group's watermark data handle;
Then, system with profit in a like fashion, once more to the inner embed watermark synchronous head signal of follow-up protected voice signal S
1(Signal), data content such as digital finger-print information, thereby improve the algorithm redundance, strengthen the anti-attack ability of watermark data;
System will follow and not carry out above-mentioned watermark information telescopiny, stop protection until the user and handle;
When the user need stop the protection processing, system will utilize that one-way hash function calculates in the whole section protected voice signal, the eap-message digest H (Data) of all group's watermark datas, and this summary info is embedded in the end of protected voice signal together with end mark End, witness marker Token, handle to finish the voice protection;
The watermark extracting end need carry out signal integrity verification to certain section shielded voice signal, or when need differentiating the attaching information of voice signal, the watermark extracting system will be according to the extraction anchor point that each synchronous head signal provided in the voice signal, protected voice signal is carried out watermark extraction process, recover watermark information Data "=Re (Signal ");
System recovers by decoding processing after all original watermark data, and the user can clearly declare the attaching information of knowing certain section voice signal according to the digital finger-print information that is comprised in each group data;
Subsequently, be masked as terminating point, be the segmentation foundation with End with the Token in the watermark data, the eap-message digest of all watermark datas in the compute segment, and if this value compared with the digest value that extracts from voice signal---identical, prove that then the related voice signal is complete; If different, prove that then the related voice signal is under attack, and can clearly declare by the little group # among the Token and to know which signal segment is under attack; (group is numbered the continuity numbering, and the user can judge which signal segment has been subjected to attack by the number of disappearance);
If the watermark extracting end does not recover the End sign, prove that then the related voice signal has excalation; The watermark extracting end needs according to the distance between a protected voice signal subsequent synchronisation signal and current, and the Token in the follow-up watermark data declares and knows whether End loses.
If the watermark extracting end fails to recover eap-message digest from voice signal, the distortion to some extent of related voice signal waveform then is described, the user can further judge its reason according to the watermark information that has extracted.
This method at first improves the redundance of watermark information by the mode that embeds synchronous head signal and digital fingerprint data repeatedly, make losing not of indivedual sound bites handle and bring tremendous influence the extraction of watermark data, thereby improved the anti-shearing attacking ability of holistic approach, guaranteed the correct extraction ratio of higher finger print information; Simultaneously this method has been introduced treatment measures such as " timestamp " and one-way hash function, provides technical guarantee for judging that watermark data is whether complete, and then also for how to verify that the integrality of voice signal has proposed the technology solution route.
The digital coding disposal route that is adopted in the technology implementation method that the present invention proposes can significantly improve the error code correction ability of holistic approach; The watermark extracting method for synchronous that the present invention proposes can successfully be realized the watermark extracting synchronized positioning under stronger interchannel noise influence; The heavy synchronous processing that is adopted in the leaching process can be corrected because signal attenuation, watermark extracting synchronized positioning mistake that signal time delay brought; In the voice signal completeness protection method that the present invention proposes based on digital finger-print information; at first utilize the mode that embeds digital finger-print information repeatedly to strengthen the anti-shearing attacking ability of holistic approach; improved the correct extraction ratio of digital finger-print information; utilize one-way hash function that watermark data in the section is calculated the mode of eap-message digest afterwards again, whether, carrier signal whether complete for watermark extracting end checking watermark information the complete scientific approach that provides.
Description of drawings
Fig. 1 is overall design approach implementing procedure figure of the present invention.
Fig. 2 is a watermark data decoding method design drawing of the present invention.
Fig. 3 is a watermarking algorithm implementation method design drawing of the present invention.
Fig. 4 is audio data stream of the present invention and watermark data stream structural representation.
Embodiment
The digital watermark technology implementation method that the present invention proposes is applicable to the embedded hardware platform, particularly based on the Embedded Real-Time voice signal processing platform of dsp chip.The watermark data Processing Algorithm that this method adopted, watermark embed that extraction algorithm, the complete verification algorithm of voice signal have all that algorithm principle is easy, fast operation, hardware resource is required advantages such as lower.The present invention utilizes DSP embedded speech signal processing platform, according to previously described design proposal, has successfully developed the digital digital watermark application apparatus of a cover.DSP embedded hardware platform of the present invention has certain programmability and data storage function, and is furnished with audio signal sample module and USB module, by the USB module can realize and PC between information interaction.In order more clearly to describe the specific embodiment of the present invention, Fig. 1 has provided the overall design approach implementing procedure:
Utilize the DSP hardware device that voice signal is protected before the processing, the operator must at first access to the DSP devices in series in original communication system, so that equipment is intercepted and captured primary speech signal in original communication system, and then embed watermark data Data forms protection to signal;
Subsequently, the operator must utilize PC that original watermark data Data such as digital finger-print information are downloaded on the data storage device in the DSP device by USB interface;
Afterwards, the DSP hardware unit will carry out digital coding to original watermark data Data to be handled, thereby obtains watermark information Data ' to be embedded=I{BCH[CRC (Data)] };
After both sides set up normal voice communication, the operator can start the voice signal protection at any time and handle, watermark information Data ' to be embedded is embedded in the protected voice signal in real time, thereby produce and carry the sound signal stream of watermark information, and utilize original communication facilities that Signal ' is sent to the speech take over party;
Implement watermark and embed when handling, the DSP equipment of watermark built-in end will cushion a certain amount of audio frequency sampling point data, and utilize the voice signal Processing Algorithm to generate synchronous head signal S
1(Signal); Afterwards, start watermark embedding processing immediately watermark data Data ' to be embedded is embedded in the primary speech signal, obtain Signal '=Em (Signal, Data ');
Watermark data stream entrained among the audio data stream Signal ' is made up of a series of packet data package, and its structure as shown in Figure 4; Wherein Pack is the interleaved encoded data bag, and Group is the BCH packet data package, and End is the watermark data stream end mark in the BCH packet data package, and each is made up of original watermark data Data and CRC cyclic redundancy check (CRC) code thereof without the packet data package of Bose-Chaudhuri-Hocquenghem Code; Each Pack packet contains one group of complete speech digit finger print information, " timestamp " and witness marker Token;
A Pack includes three BCH packet data package among Fig. 4, and this structure can be adjusted (can increase BCH packet number in right amount) at any time according to application demand; The introducing of Pack grouping bag is in order to increase interleave depth, and the continuity random error is distributed in each Bose-Chaudhuri-Hocquenghem Code group effectively, improves code correction efficient;
The watermark built-in end, DSP equipment embeds digital finger-print information repeatedly in the mode of a Pack packet of a synchronous head signal cooperation, its objective is the redundance that will improve finger print information, the anti-shearing attacking ability of enhanced system;
The watermark built-in end, when the operator cancels the protection of voice signal, DSP equipment can be handled the protection of voice signal with automatic termination with eap-message digest H (Data), the end mark End of watermark data in protected section, witness marker Token as stopping the end that packet is embedded in protected voice signal;
After the protection processing end to voice signal, the user need not to carry out operation bidirectional, and the DSP device will transit to no watermark treatment state reposefully;
When the watermark extracting end carries out watermark extraction process, DSP equipment will at first cushion N
1" audio frequency sampling point data are also searched for synchronous head signal to individual Signal, wherein N
1Should be greater than the pairing voice signal length of complete group watermark data; Detect after the synchronous head signal, the locating information that DSP equipment will provide according to this signal is to N in the buffer zone
1Individual Signal " the audio frequency sampling point carries out watermark extraction process, thereby obtains Data "=Re (Signal ");
Extract Data " after, DSP equipment will serve as that the basis splits Data with structure shown in Figure 4 " packet, and it is carried out deinterleaving handle, DI (Data ") obtained; Afterwards again according to the BCH packet configuration to DI (Data ") carries out correction process, obtain DBCH[DI (Data ")]; Utilize the most at last the CRC algorithm to DBCH[DI (Data ")] carry out verification, if CRC{DBCH[DI (Data ")] } result is correct, represents that then the watermark extracting operation is accurate, reduciblely goes out original watermark data Data;
If CRC{DBCH[DI (Data ")] } check errors, then represent with current synchronized positioning point to be that benchmark can't correctly recover original watermark data Data, the watermark extracting end needs N
1" the audio frequency sampling point weighs synchronously and adjusts individual Signal, obtains S
2(Signal "); After finishing adjustment, the watermark extracting end will utilize the watermark extracting algorithm to S once more
2(Signal ") handles, and obtains Data "=Re[S
2(Signal ")] and attempt from Data " recovery original watermark data Data;
If after reorientation was adjusted repeatedly, the watermark extracting end still can not recover original watermark data Data, represents that then this group error in data is too much, the watermark extracting operation failure;
After the watermark extracting end extracts end mark End, equipment will stop the watermark extraction process process automatically; By the digital finger-print information that is contained in each group's watermark data bag, the user can clearly declare the attaching information of knowing protected voice signal;
Equipment utilizes one-way hash function that all watermark informations within the same protected speech signal segments are calculated eap-message digest, and this digest value and the digest value that extracts is compared according to the witness marker Token in each group's watermark data subsequently; If be worth identically, illustrate that then protected voice signal is complete; If value is different, then wave form varies has taken place in the declaratives voice signal, at this moment, the user can further declare according to the little group # of the watermark data among the Token and know which signal segment has been subjected to attack (group is numbered the continuity numbering, and the user can judge which signal segment has been subjected to attack by the number of disappearance);
If the watermark extracting end does not recover the End sign, prove that then the related voice signal has excalation; The watermark extracting end needs according to the distance between a protected voice signal subsequent synchronisation signal and current, and the Token in the follow-up watermark data declares and knows whether End loses.
If the watermark extracting end fails to recover eap-message digest from voice signal, the distortion to some extent of related voice signal waveform then is described, the user can further make synthetic determination according to the watermark information that has extracted.
The watermark extracting end recovers after the original watermark data, shows but the USB interface in operator's use device is uploaded to PC with the watermark information that extracts; Equipment also can be uploaded some auxiliary judgement information (for example watermark data that does not recover fully) simultaneously, for the analysis-by-synthesis error reason provides the data support.
The present invention proposes a kind of voice integrity protection method based on digital finger-print, its characteristics are: (1) this method has proposed the comparatively perfect digital watermark application scheme of a cover around a kind of strong robustness watermarking algorithm; (2) this method has adopted the same footwork of energy jump formula watermark extracting, and the method can realize extracting synchronously location effectively under stronger noise effects; (3) in watermark extraction process, for further improving synchronous accuracy rate, this method has adopted dynamic heavy synchronization processing method, namely eliminates by the position of the synchronous anchor point of dynamic adjustment because the site drift that channel delay brings; Whether (4) this method has proposed perfect combined type watermark data coding method, comprise adopting CRC to encode to check watermark data correct, adopts Bose-Chaudhuri-Hocquenghem Code to come correcting data miss-code, adopts interweaving encoding to disperse the continuity random error; (5) this method utilization mode of repeatedly embedding group's watermark data has improved the anti-shearing attacking ability of holistic approach, utilizes simultaneously one-way hash function to come accessory system to declare and knows whether protected voice signal is complete; This method organically combines all kinds of algorithms, has realized that the engineering of digital watermark is used, and algorithm structure compactness, lower to the hardware resource requirement of implementation platform.
Claims (3)
1. voice integrity protection method based on digital finger-print.This method relies on general voice communication system; based on digital watermark technology; realized embedding digital finger-print information, this technology contents of integrity verification information in real time in communication process in voice signal, having made up is the voice signal completeness protection method of judgment basis with the digital watermark information.This method mainly is made of following three partial contents, i.e. (1) watermarking algorithm implementation method; (2) original watermark data-encoding scheme; (3) voice signal completeness protection method, the specific implementation step is as follows:
1) watermark built-in end at first carries out Error Correction of Coding to original watermark data Data and handles, and can obtain watermark information Data ' to be embedded; This coding method is the combined encoding method, and its content comprises mistake and disperses encryption algorithm (interweaving encoding algorithm), Error Correction of Coding algorithm (Bose-Chaudhuri-Hocquenghem Code algorithm) and error detection coding algorithm (CRC encryption algorithm);
2) after user's initiation is handled the protection of voice signal, the watermark built-in end will utilize speech processing algorithm to generate energy jump formula synchronous head signal S
1(Signal), and with this signal be embedded in the protected speech signal segments;
3) subsequently, the watermark built-in end starts watermark immediately and embeds processing, and watermark information Data ' to be embedded is embedded in the protected voice signal, obtains Signal '=Em (Signal, Data '), handles with the embedding of finishing group's watermark data;
4) the watermark built-in end constantly repeats the telescopiny (in each group's watermark data, digital finger-print information is identical) of above-mentioned synchronous head signal and group's watermark data, until the protection of user's termination to voice signal;
5) when user's termination is handled the protection of voice signal, the watermark built-in end will utilize one-way hash function to calculate the eap-message digest of all watermark datas in the protected speech signal segments, and this digest value is embedded in the end of protected voice signal together with end mark End, witness marker Token as last group watermark data, finish protection and handle;
6) have certain waveform loss and be mingled with some interchannel noises after the Signal ' process channel voice compression coding, so the signal that the watermark extracting end obtains is Signal ";
When 7) user need carry out integrity verification to voice signal, the watermark extracting end will start watermark information immediately and extract processing; At first extract end and detect after the synchronous head signal in the voice signal, the locating information that will provide according to this signal is to Signal " carry out watermark extraction process, can get Data "=Re (Signal ");
8) afterwards, the watermark extracting end is to Data " carry out decoding processing, if CRC{DBCH[DI (Data ")] } check results is correct, then can recover original watermark data Data; If CRC{DBCH[DI (Data ")] } the check results mistake, then the watermark extracting end must utilize heavy synchronized algorithm processes voice signals Signal ", can get S
2(Signal ");
9) if Data "=Re[S is arranged
2(Signal ")], and CRC{DBCH[DI (Data ")] } check results is correct, then represents can recover original watermark data Data by heavy synchronous adjustment; If through still recovering the original watermark data after the heavy synchronous processing repeatedly, illustrate that then this group watermark data bit error rate is too high, the watermark extracting failure;
10) the watermark extracting end will be carried out above-mentioned watermark extraction process process repeatedly until extracting end mark End, and the user can judge the attaching information of voice signal by the digital finger-print in any group watermark data;
11) subsequently, the watermark extracting end will be numbered the segmentation foundation with the Token of each group's number in then, the eap-message digest of all watermark datas in the compute segment, and this digest value and the digest value that extracts compared; If be worth identically, illustrate that then protected voice signal is complete; If value is different, the distortion to some extent of related voice signal waveform then is described, the user can further declare according to the little group # of the watermark data among the Token and know which signal segment has been subjected to attack (group is numbered the continuity numbering, and the user can judge which signal segment has been subjected to attack by the number of disappearance);
12) if the watermark extracting end does not recover the End sign, then proof ending voice signal waveform distortion to some extent; The watermark extracting end needs according to the distance between a protected voice signal subsequent synchronisation signal and current, and the Token in the follow-up watermark data declares and knows whether End loses;
13) if the watermark extracting end fails to recover eap-message digest from voice signal, the distortion to some extent of related voice signal waveform is described then, the user can further judge its reason according to the watermark information that has extracted.
Before original watermark information Data is carried out encoding process, must require Data is carried out packet according to Bose-Chaudhuri-Hocquenghem Code in the step 1), the data length that should satisfy CRC (Data) meets selected Bose-Chaudhuri-Hocquenghem Code length; For improving the wrong dispersion efficiency of interweaving encoding in the step 1), a plurality of BCH packet data package must be combined and carry out interleaving treatment.Step 2) the energy synchronous head signal in is a high energy signal, the watermark extracting end remains with identical sample of signal, when system carries out the synchronous head signal detection, this signal and the synchronous head signal that receives can be done auto-correlation processing, the correlativity highest point just is the synchronized positioning point.The principle of heavy synchronized algorithm is dynamically to adjust the watermark extracting anchor point in time domain in the step 8), and wherein adjusting step-length and setting range can be set up on their own according to practical application request by the user.
2. according to the original watermark data-encoding scheme described in the claim 1, it is characterized in that: coding method has comprised the coding techniques of three class maturations, promptly be used for distributed channel continuity burst random error interweaving encoding, be used for error correction the linear group coding of BCH, be used to check wrong CRC cyclic redundancy check (CRC) code; The design concept of method can be expressed as at first and by interweaving encoding the continuity burst error is distributed in each BCH data set to improve the error correction efficient of coding method, the mistake of utilizing Bose-Chaudhuri-Hocquenghem Code to correct in each data set is again transmitted accuracy to improve, finally utilize the correctness of CRC code detection data, this group coding algorithm further comprises following steps:
1) original watermark information Data is carried out packet, and obtain CRC (Data)=Data+CRC by the CRC coding, wherein CRC represents the check code of Data;
2) handle CRC (Data) by Bose-Chaudhuri-Hocquenghem Code and obtain BCH[CRC (Data)];
3) interweaving encoding is carried out in three groups of BCH data combinations together obtains Data '=I{BCH[CRC (Data)] }, thus the combined encoding processing procedure of watermark data finished.
3. according to the voice signal completeness protection method described in the claim 1, it is characterized in that: utilize one-way hash function that all watermark informations in the protected voice signal are handled and obtain eap-message digest H (Data), and this summary info is embedded in the end of protected voice signal; Whether the user is complete according to the watermark data carrier signal whether complete, that can verify watermark data that summary data H (Data) decidable that recovers extracts; Simultaneously, the user can further declare according to the witness marker Token in the watermark data and know which sound bite has been subjected to attack.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010132256XA CN101847409B (en) | 2010-03-25 | 2010-03-25 | Voice integrity protection method based on digital fingerprint |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010132256XA CN101847409B (en) | 2010-03-25 | 2010-03-25 | Voice integrity protection method based on digital fingerprint |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101847409A true CN101847409A (en) | 2010-09-29 |
| CN101847409B CN101847409B (en) | 2012-01-25 |
Family
ID=42772004
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010132256XA Expired - Fee Related CN101847409B (en) | 2010-03-25 | 2010-03-25 | Voice integrity protection method based on digital fingerprint |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101847409B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013053224A1 (en) * | 2011-10-09 | 2013-04-18 | 华为技术有限公司 | Data transmission method and sending end and receiving end |
| CN103514876A (en) * | 2012-06-28 | 2014-01-15 | 腾讯科技（深圳）有限公司 | Method and device for eliminating noise and mobile terminal |
| CN105681271A (en) * | 2014-12-08 | 2016-06-15 | 三星电子株式会社 | Method and apparatus for providing integrity check data |
| CN106997388A (en) * | 2017-03-30 | 2017-08-01 | 宁波亿拍客网络科技有限公司 | A kind of image and non-image labeling method, equipment and application process |
| CN107293306A (en) * | 2017-06-21 | 2017-10-24 | 湖南省计量检测研究院 | A kind of appraisal procedure of the Objective speech quality based on output |
| CN107666326A (en) * | 2017-10-23 | 2018-02-06 | 东南大学 | A kind of joint BCH and CRC Polarization Coding, coding/decoding method and device |
| CN108650054A (en) * | 2018-04-03 | 2018-10-12 | 厦门大学 | The method for establishing private communication channel with the network flow watermark of forward error correction and interlaced code |
| CN110896503A (en) * | 2018-09-13 | 2020-03-20 | 浙江广播电视集团 | Video and audio synchronization monitoring method and system and video and audio broadcasting system |
| CN111292756A (en) * | 2020-01-19 | 2020-06-16 | 成都嗨翻屋科技有限公司 | Compression-resistant audio silent watermark embedding and extracting method and system |
| CN111757168A (en) * | 2019-03-29 | 2020-10-09 | 腾讯科技（深圳）有限公司 | Audio decoding method, device, storage medium and equipment |
| CN114120466A (en) * | 2021-11-22 | 2022-03-01 | 浙江嘉科电子有限公司 | Coding and decoding device and method for polling information exchange |
| CN115174601A (en) * | 2022-06-23 | 2022-10-11 | 中国工商银行股份有限公司 | Data processing method, system, processor and electronic equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090199008A1 (en) * | 2002-11-27 | 2009-08-06 | Petar Horvatic | Watermarking digital representations that have undergone lossy compression |
| WO2009144470A1 (en) * | 2008-05-29 | 2009-12-03 | Intrasonics S.à.r.l. | Data embedding system |
| CN101645162A (en) * | 2009-09-09 | 2010-02-10 | 北京邮电大学 | Printing-scanning resistance zero-watermarking algorithm |
-
2010
- 2010-03-25 CN CN201010132256XA patent/CN101847409B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090199008A1 (en) * | 2002-11-27 | 2009-08-06 | Petar Horvatic | Watermarking digital representations that have undergone lossy compression |
| WO2009144470A1 (en) * | 2008-05-29 | 2009-12-03 | Intrasonics S.à.r.l. | Data embedding system |
| CN101645162A (en) * | 2009-09-09 | 2010-02-10 | 北京邮电大学 | Printing-scanning resistance zero-watermarking algorithm |
Non-Patent Citations (1)
| Title |
|---|
| 《遥测遥控》 20071231 白婷等 BCH(31,16)、CRC-32与交织码级联的实现及分析 第28卷, 第6期 2 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013053224A1 (en) * | 2011-10-09 | 2013-04-18 | 华为技术有限公司 | Data transmission method and sending end and receiving end |
| CN103514876A (en) * | 2012-06-28 | 2014-01-15 | 腾讯科技（深圳）有限公司 | Method and device for eliminating noise and mobile terminal |
| CN105681271B (en) * | 2014-12-08 | 2019-03-29 | 三星电子株式会社 | The method and apparatus of integrity checking data is provided |
| CN105681271A (en) * | 2014-12-08 | 2016-06-15 | 三星电子株式会社 | Method and apparatus for providing integrity check data |
| US10516677B2 (en) | 2014-12-08 | 2019-12-24 | Samsung Electronics Co., Ltd. | Method and apparatus for providing integrity check data |
| CN106997388A (en) * | 2017-03-30 | 2017-08-01 | 宁波亿拍客网络科技有限公司 | A kind of image and non-image labeling method, equipment and application process |
| CN107293306A (en) * | 2017-06-21 | 2017-10-24 | 湖南省计量检测研究院 | A kind of appraisal procedure of the Objective speech quality based on output |
| CN107293306B (en) * | 2017-06-21 | 2018-06-15 | 湖南省计量检测研究院 | A kind of appraisal procedure of the Objective speech quality based on output |
| CN107666326A (en) * | 2017-10-23 | 2018-02-06 | 东南大学 | A kind of joint BCH and CRC Polarization Coding, coding/decoding method and device |
| CN107666326B (en) * | 2017-10-23 | 2020-10-02 | 东南大学 | Polarization coding and decoding method and device combining BCH and CRC |
| CN108650054A (en) * | 2018-04-03 | 2018-10-12 | 厦门大学 | The method for establishing private communication channel with the network flow watermark of forward error correction and interlaced code |
| CN108650054B (en) * | 2018-04-03 | 2020-06-09 | 厦门大学 | Method for establishing concealed channel by network stream watermark of forward error correcting code and interleaving code |
| CN110896503A (en) * | 2018-09-13 | 2020-03-20 | 浙江广播电视集团 | Video and audio synchronization monitoring method and system and video and audio broadcasting system |
| CN111757168A (en) * | 2019-03-29 | 2020-10-09 | 腾讯科技（深圳）有限公司 | Audio decoding method, device, storage medium and equipment |
| CN111292756A (en) * | 2020-01-19 | 2020-06-16 | 成都嗨翻屋科技有限公司 | Compression-resistant audio silent watermark embedding and extracting method and system |
| CN111292756B (en) * | 2020-01-19 | 2023-05-26 | 成都潜在人工智能科技有限公司 | Compression-resistant audio silent watermark embedding and extracting method and system |
| CN114120466A (en) * | 2021-11-22 | 2022-03-01 | 浙江嘉科电子有限公司 | Coding and decoding device and method for polling information exchange |
| CN115174601A (en) * | 2022-06-23 | 2022-10-11 | 中国工商银行股份有限公司 | Data processing method, system, processor and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101847409B (en) | 2012-01-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101847409B (en) | Voice integrity protection method based on digital fingerprint | |
| CN102074240B (en) | Digital audio watermarking algorithm for copyright management | |
| CN101512638B (en) | Method and apparatus for encoding/decoding symbols carrying payload data for watermarking of an audio or video signal | |
| CN101271690B (en) | Audio spread-spectrum watermark processing method for protecting audio data | |
| CN103024244B (en) | Detector, the method and computer program of the watermark during the embedded device in being represented to information by watermark embedment, Detection Information represent | |
| CN103368660B (en) | A kind of bionical underwater acoustic communication method based on difference Pattern pattern time-delay coding | |
| CN101645743B (en) | Chaotic sequence-based packet M element spread spectrum communication method and device | |
| MX2009012385A (en) | Apparatus and method for encoding and decoding signals. | |
| CN101206861B (en) | Method for imbedding digital music production authentication information and method for authentication of said production | |
| CN101918999A (en) | Methods and apparatus to perform audio watermarking and watermark detection and extraction | |
| CN1054693A (en) | The continous cipher of cellular communication system is synchronous | |
| GB2432764A (en) | Method and apparatus of turbo encoder | |
| DE60326578D1 (en) | REINTERBATION OF WATERMARK IN MULTIMEDIA SIGNALS | |
| CN103346830A (en) | Voice transmission method and device based on Beidou satellite | |
| CN106448686B (en) | A kind of adaptive audio airspace steganography method | |
| CN110163787A (en) | Digital audio Robust Blind Watermarking Scheme embedding grammar based on dual-tree complex wavelet transform | |
| CN102624518A (en) | Voice encryption and decryption method | |
| GB0418279D0 (en) | System for providing access to operation information | |
| WO2003056557A3 (en) | Method, system, and program for synchronization and resynchronization of a data stream | |
| WO2006108940A3 (en) | Method for communicating between a reader and a wireless identification marker, associated reader and marker | |
| WO2004079911A3 (en) | Clock and data recovery method and apparatus | |
| CN1889428A (en) | Biological characteristic data safety hidden transmitting method under network environment | |
| CN1331144C (en) | Embedding first digital information signal into a second digital information signal for transmission via a transmission medium | |
| CN101458929A (en) | Generating water marked copies of audio signals and detecting them using a shuffle data store | |
| CN101778289B (en) | Method for embedding and extracting digital video information based on FLV (flash video) video structural feature |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20120125 Termination date: 20130325 |