CN115618397B - Voice encryption method for recording pen - Google Patents
Voice encryption method for recording pen Download PDFInfo
- Publication number
- CN115618397B CN115618397B CN202211629107.3A CN202211629107A CN115618397B CN 115618397 B CN115618397 B CN 115618397B CN 202211629107 A CN202211629107 A CN 202211629107A CN 115618397 B CN115618397 B CN 115618397B
- Authority
- CN
- China
- Prior art keywords
- sequence
- vector
- dictionary
- recording
- sparse
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 239000013598 vector Substances 0.000 claims abstract description 299
- 239000011159 matrix material Substances 0.000 claims abstract description 69
- 230000009467 reduction Effects 0.000 claims description 5
- 230000005540 biological transmission Effects 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 239000000126 substance Substances 0.000 description 3
- 230000000739 chaotic effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000005336 cracking Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 1
- 239000010931 gold Substances 0.000 description 1
- 229910052737 gold Inorganic materials 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000003607 modifier Substances 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/602—Providing cryptographic facilities or services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/237—Lexical tools
- G06F40/242—Dictionaries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/27—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Engineering & Computer Science (AREA)
- Computer Security & Cryptography (AREA)
- Artificial Intelligence (AREA)
- Signal Processing (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Bioethics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Storage Device Security (AREA)
Abstract
The invention relates to the field of secure transmission, in particular to a voice encryption method for a recording pen, which acquires a recording sequence to obtain a recording sequence set; obtaining a dictionary matrix and a plurality of sparse vectors according to the recording sequence set; obtaining a topic feature vector according to the dictionary matrix and the sparse vector; obtaining a projection value of each dictionary vector according to the topic feature vector, obtaining a first similarity of each dictionary vector according to the recording sequence, the sparse vector and the dictionary matrix, and obtaining the generalization of each dictionary vector according to the first similarity and the projection value of each dictionary vector; obtaining the updated data number of each dictionary vector according to the generalization of each dictionary vector, and obtaining an updated dictionary matrix according to the updated data number of each dictionary vector and the key sequence; and obtaining ciphertext data of each recording sequence according to the updated dictionary matrix and the sparse vector, thereby realizing different encryption of the recording sequences according to the importance of each information in the recording sequences and ensuring the safety of the recording sequences.
Description
Technical Field
The application relates to the field of encryption transmission, in particular to a voice encryption method for a recording pen.
Background
With the development of society, the privacy protection of people is increasingly enhanced, and recording is used as an important work and life recording means, can record aspects of work and life, can provide convenience for work and life of people to a great extent when the recording is reasonably utilized, but records some private information after all, and when the private information is stolen by a minded person, the private information is easily used for illegal profit making, so that the acquired recording needs to be encrypted and protected.
When the traditional encryption method is used for encryption, the characteristic of information to be protected is not considered, a certain theme generally exists for the recording information, the recording content can be expanded around the theme, the subject word is used as the important content for bearing the recording information and can repeatedly appear in the recording, so that the subject word of the recording can be found according to the characteristic, the subject word is used as the central core content of the recording information and is important for the recording information, and the information related to the subject word needs to be protected in a key way when the recording information is encrypted. Meanwhile, not all information in the recording information is important for mastering the recording content, for example, information such as adjectives with personal colors in the recording has small influence on mastering the recording content information, the general information capable of describing the recording content in the recording is important, the general information serving as the recording content generally has large correlation with a subject word, the general information has large correlation with structural information of the recording, and the general information is important for the recording information as core content in the recording information, so that the information needs to be protected in a key mode when recording encryption is carried out. Therefore, different information in the recording is encrypted and protected differently, so that the recording content is effectively protected, and the encryption efficiency can be improved.
Disclosure of Invention
In order to solve the technical problem, the invention provides a voice encryption method for a recording pen, which comprises the following steps:
acquiring a recording sequence to obtain a recording sequence set;
obtaining a dictionary matrix and a plurality of sparse vectors according to the recording sequence set; the dictionary matrix comprises a plurality of dictionary vectors, each element in each sparse vector is called as a sparse value, and one sparse value in each sparse vector corresponds to one dictionary vector;
obtaining a difference sequence of each sparse value according to the relation between the dictionary matrix, the sparse values and the recording sequence information; obtaining the theme of each dictionary vector according to the condition that the difference sequence of each sparse value has the theme word characteristics, and taking the dictionary vector with the maximum theme as a theme characteristic vector; taking the projection value of each dictionary vector in the topic feature vector as the projection value of each dictionary vector, and obtaining the first similarity of each dictionary vector according to the similarity of the difference sequence of each sparse value and the structural information of the recording sequence; obtaining the generalization of each dictionary vector according to the first similarity and the projection value of each dictionary vector;
obtaining the updated data number of each dictionary vector according to the generalization of each dictionary vector, obtaining a position sequence and a first key sequence according to the updated data number and the key sequence of each dictionary vector, and obtaining an updated dictionary matrix according to the updated data number, the position sequence and the first key sequence of each dictionary vector; and obtaining the ciphertext data of each recording sequence according to the updated dictionary matrix and the sparse vector.
Preferably, the obtaining of the difference sequence of each sparse value according to the relationship between the dictionary matrix, the sparse value and the recording sequence information includes the specific steps of:
for any sparse vector, acquiring any one sparse value in the sparse vectors, recording the sparse value as a target sparse value, adjusting the target sparse value to 0, keeping other sparse values unchanged to obtain an adjusted sparse vector of the target sparse value, and obtaining an adjusted recording sequence of the target sparse value according to the dictionary matrix and the adjusted sparse vector of the target sparse value;
calculating the difference value of the recording sequence and the adjusted target sparse value to obtain a difference value sequence of the target sparse value;
and obtaining a difference value sequence of each sparse value in all the sparse vectors.
Preferably, the obtaining the theme of each dictionary vector according to the condition that the difference sequence of each sparse value has the feature of the subject word includes the specific steps of:
dividing continuous non-0 elements in each difference value sequence of each sparse value into a non-zero sequence segment to obtain a plurality of non-zero sequence segments; and calculating the matching value of each non-zero sequence segment and other non-zero sequence segments, taking the mean value of the matching values of each non-zero sequence segment and all other non-zero sequence segments as the comprehensive matching value of each non-zero sequence segment, and taking the mean value of the comprehensive matching values of all non-zero sequence segments as the theme of each dictionary vector.
Preferably, the obtaining of the first similarity of each dictionary vector according to the similarity between the difference sequence of each sparse value and the recording sequence structure information includes the specific steps of:
and downsampling each recording sequence to obtain a recording sequence after dimensionality reduction, and calculating the similarity between the difference sequence of each sparse value and the recording sequence after dimensionality reduction to obtain the first similarity of each dictionary vector.
Preferably, the obtaining of the updated dictionary matrix according to the updated data number, the position sequence and the first key sequence of each dictionary vector includes the following specific steps:
setting the mark position to one; recording a dictionary vector of a marked position in a dictionary vector sequence as a target vector, recording the number of elements contained in the target vector as a first number, and obtaining an updated dictionary vector of the target vector according to the target vector, the position sequence and a first key sequence, wherein the method comprises the following steps:
acquiring dictionary vectors between a first position in the dictionary vector sequence and a position before the target vector to obtain a dictionary vector set, and obtaining a plurality of second positions according to the number of updated data of each dictionary vector in the dictionary vector set; acquiring data at a plurality of second positions in the position sequence to obtain a plurality of first data; taking the first data and the remainder of the first number as an updating position to obtain a plurality of updating positions; acquiring data at a plurality of updating positions in a target vector to obtain a plurality of data to be updated, acquiring data at a plurality of second positions in a first key sequence to obtain a plurality of intermediate data, acquiring a plurality of replacement data according to the plurality of intermediate data and the plurality of data to be updated, and replacing the plurality of data to be updated in the target vector by using the corresponding replacement data respectively to obtain an updated dictionary vector of the target vector;
adding one to the mark position, marking the dictionary vector at the mark position in the dictionary vector sequence as a target vector, and repeatedly executing the following steps: and obtaining an updated dictionary vector of the target vector according to the target vector, the position sequence and the first key sequence, and ending when the mark position is more than or equal to the number of elements of the dictionary vector to obtain an updated dictionary matrix.
Preferably, the obtaining of the ciphertext data of each recording sequence according to the updated dictionary matrix and the sparse vector includes the following specific steps:
wherein the content of the first and second substances,denotes the firstA number of sparse vectors, each of which is a sparse vector,representing the updated dictionary matrix after the update of the dictionary matrix,denotes the firstCiphertext data of the recording sequence.
Preferably, the obtaining of the position sequence and the first key sequence according to the number of update data of each dictionary vector and the key sequence includes the specific steps of:
calculating the sum of the updated data numbers of all dictionary vectors to obtain a first number, and intercepting the first position in the key sequence toThe sequence at a position is denoted as the sequence of positions,taking a first number; the position sequence is cut out from the key sequence to obtain a first key sequence.
The embodiment of the invention at least has the following beneficial effects: the general information of the recording sequence is helpful for rapidly grasping the recording sequence information, so when the recording sequence is encrypted, the general information in the recording sequence needs to be encrypted in a complex way, so that the safety of the general information is ensured, when the general property of each dictionary vector is analyzed, the dictionary vector with the general characteristic is considered to be related to the subject word of the recording sequence, and meanwhile, the general property of each dictionary vector is obtained based on the characteristic similar to the structural information of the recording sequence, so that the dictionary vectors are encrypted differently according to the general property of each dictionary vector.
The method comprises the steps of obtaining the number of updating data of each dictionary vector according to the generalization of each dictionary vector, obtaining a position sequence and a first key sequence according to the key sequence and the number of updating data of each dictionary vector, obtaining an updated dictionary matrix according to the number of updating data, the position sequence and the first key sequence of each dictionary vector, and obtaining ciphertext data of each recording sequence according to the updated dictionary matrix and the sparse vector.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart of a voice encryption method for a recording pen according to the present invention;
FIG. 2 shows a voice encryption method for a recording pen according to the present inventionAn algorithm structure schematic diagram;
fig. 3 is a schematic diagram of a dictionary matrix updating method of a voice encryption method of a recording pen according to the present invention.
Detailed Description
To further illustrate the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description of the voice encryption method for a recording pen according to the present invention, the specific implementation, structure, features and effects thereof will be given with reference to the accompanying drawings and the preferred embodiments. In the following description, different "one embodiment" or "another embodiment" refers to not necessarily the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following describes a specific scheme of the voice encryption method for the recording pen in detail with reference to the accompanying drawings.
Referring to fig. 1, a flowchart illustrating steps of a voice encryption method for a recording pen according to an embodiment of the present invention is shown, where the method includes the following steps:
and S001, acquiring a recording sequence set and generating a key sequence.
1. Acquiring a recording sequence set:
in order to prevent the leakage of the recording information collected by the recording pen and the loss of the interest of the party, the recording information needs to be encrypted, so that the recording information needs to be collected.
The method comprises the steps of acquiring a plurality of recording sequences by using a recording pen, wherein the plurality of recording sequences form a recording sequence set, and it needs to be stated that each recording sequence corresponds to a complete recording theme.
Uploading the recording sequence set to a server, and carrying out calculation analysis on the recording sequence set through the server to realize encryption of the recording sequence.
2. Generating a key sequence:
generating a length ofChaotic sequence of (1)The chaos sequence is called as a key sequence, and the hyper-parameters of the chaos mapping function are agreed by an encryption party and a decryption party in advance without transmission.
Each data in the key sequence is referred to as a key value.
Step S002, calculating the projection value and the first similarity of each dictionary vector, and obtaining the generalization of each dictionary vector according to the projection value and the first similarity of each dictionary vector.
Because the importance of each information in the recording sequence is different, some information is helpful to quickly grasp the recording content, such as the subject word information in the recording sequence and the general information in the recording sequence, and some information has little influence on grasping the recording content, such as some modifier word information in the recording sequence, so the importance of different information in the recording information on grasping the recording content needs to be analyzed to obtain the recording information.
Meanwhile, when the recording sequence is encrypted, the recording sequence data is directly encrypted, actual contents in the recording sequence cannot be completely concealed, particularly information with obvious statistical characteristics in the recording sequence, the recording sequence has a series of generalized characteristics, the characteristic information can better conceal main information in the recording sequence, if the characteristic information of the recording sequence is encrypted, the characteristic attribute of the recording sequence can be directly changed at a low cost, so that the contents of the recording sequence can be better concealed, and the K-SVD algorithm is used as a compression algorithm and can better extract the characteristic information of the recording sequence, so that the characteristic information in the recording sequence can be extracted by using the K-SVD algorithm and then encrypted.
Acquiring a dictionary matrix and a sparse vector of the recording sequence:
taking a plurality of recording sequences in the recording sequence set as input, and performing learning training by using a K-SVD algorithm to obtain a dictionary matrix and a plurality of sparse vectors; one recording sequence corresponding to one sparse vectorPersonal recordingSequence is described asCorresponding to itSparse vector is notedThe dictionary matrix is recorded asOf 1 atA recording sequence andthe sparse vectors satisfy the following relationship:
first, theA recording sequenceIs of length ofSequence of (1) ARecording sequenceTo middleA data note asThe dictionary matrix isMatrix of dimensions, the first in the dictionary matrixGo to the firstData of the column is recorded asTo be connected toA first in a sparse vectorRecording the data as sparse valuesThus, the equation can be expressed as:
as can be seen from the above equation, the first in the sparse vectorThe first in the personal data and dictionary matrixThe columns have a corresponding relationship, that is, each sparse value of each sparse vector corresponds to a column of data in the dictionary matrix, and each column of data in the dictionary matrix is recorded as a dictionary vector, that is, the nth column represents the nth dictionary vector and also represents a feature of the audio sequence, as shown in fig. 2, fig. 2 is a diagram of a method for generating a dictionary by using a dictionary matrix, where each column of data in the dictionary matrix corresponds to a column of data in the dictionary matrix, and each column of data in the dictionary matrix represents a feature of the audio sequence, and each column of data in the dictionary matrix represents an nth dictionary vector, that is, the nth column represents an nth dictionary vector, and fig. 2 shows a feature of the audio sequenceSchematic of the algorithm, FIG. 2Each of the sparse values of the sparse vector is represented separately,representing dictionary vectors in a dictionary matrix separately, wherein the sparse valuesCorresponding to dictionary vectorsSparse valueCorresponding dictionary vectorAnalogy in sequence, \ 8230;, sparse valueCorresponding dictionary vector. The dictionary matrix obtains all the characteristics by learning a plurality of recording sequences, each dictionary vector describes one characteristic, and each sparse value in the sparse vector corresponding to each recording sequence is the condition that each recording sequence has corresponding characteristics.
Number of lines in the dictionary matrix in this embodimentGet 800, column numberTake 1200, length of key sequence。
Calculating the themes of each dictionary vector:
since the information in the recording sequence is developed based on the subject word, and the subject word generally appears repeatedly for many times, when the dictionary vector describes the feature information of the subject word, the information in the recording sequence corresponding to the feature information has the feature appearing repeatedly, and the theme of each dictionary vector is determined based on the feature.
Will be firstA first of the sparse vectorsOne sparse value is adjusted to 0 and the other sparse values are left unchanged with respect to the second sparse valueA first of sparse valuesAdjusted sparse vectorAt this time, the first stepA first of the sparse vectorsRemoving the feature corresponding to the sparse value, wherein the feature is the second in the dictionary matrixDictionary vector description, dictionary matrixAnd with respect toA number of sparse valuesAdjusted sparse vectorDot product ofA number of sparse valuesAn adjusted recording sequenceThe recorded sequence does not containFeatures described by a dictionary vector.
In summary, it is to be understood that,is reflected by the recording after the nth characteristic in the recording sequence is removed, andreflecting a recording without any features removed.
Will be firstA recording sequenceAnd with respect toA first of sparse valuesAn adjusted recording sequenceMake a difference to obtain aA number of sparse valuesA sequence of difference valuesThe information contained in the difference sequence is the first information in the recording sequenceThe dictionary vector describes information corresponding to the features.
Will relate toA number of sparse valuesA sequence of difference valuesThe continuous and uninterrupted non-0 element in (1) is divided into a non-zero sequence segment to obtain a plurality of non-zero sequence segments, it should be noted that the continuous and uninterrupted non-0 element means five or more non-0 uninterrupted elements, the matching value of each non-zero sequence segment and other non-zero sequence segments is calculated by using a DTW algorithm, the matching values of each non-zero sequence segment and all other non-zero sequence segments are averaged to obtain the comprehensive matching value of each non-zero sequence segment, the comprehensive matching values of all non-zero sequence segments are averaged, and the average value is used as the average value of the second non-zero sequence segmentA first of sparse valuesA sequence of difference valuesSubject matter ofThe larger the value is, the more the description isIn a sparse vectorThe probability that the characteristic information corresponding to the sparse value is the theme information is high, so that when the characteristic information is leaked, a thief can easily master the recording content, and the more important the characteristic information is. Due to the relation toA first of sparse valuesA sequence of difference valuesThe information of (A) is the information of the recording sequenceThe dictionary vector describes information corresponding to the feature, and thusIs also the firstThe themes of dictionary vectors.
Obtaining a dictionary vector corresponding to the maximum thematic value, wherein the dictionary vector has more information describing the thematic words, and the dictionary vector is used as a thematic feature vector。
Calculate the generalization of each dictionary vector:
since the general information of the recorded sequence is expanded around the subject word information, the general information of the recorded sequence has high correlation with the subject word information, and when the dictionary vector describes the feature corresponding to the general information of the recorded sequence, the dictionary vector has high correlation with the subject feature vector, so that the possible dictionary vector with the general feature is obtained based on the above, specifically, the following steps are carried out:
when the relevance of the dictionary vector and the topic feature vector is high, the projection value of the dictionary vector on the topic feature vector should be large. Using the following formulaThe projection value of a dictionary vector on a topic feature vector is:
wherein the content of the first and second substances,is a topic feature vector, which mainly describes feature information related to a topic word,is shown asA vector of the dictionary is generated by the dictionary vector,the length of the modulus of the vector is represented,is shown asThe projected values of the dictionary vectors.
Using Gauss gold pyramid algorithmA recording sequenceFive times of filtering and five times of downsampling processing are carried out to obtain the recording sequence after dimension reductionThe recording sequence is the structural information of the recording sequence, so that the recording sequence after dimension reduction is the structural information of the recording sequence.
Will be firstA first of the sparse vectorsOne sparse value is adjusted to 0 and the other sparse values are left unchanged with respect to the second sparse valueA first of sparse valuesAdjusted sparse vectorAt this time, the first stepA first of sparse vectorsRemoving the feature corresponding to the sparse value by the number one in the dictionary matrixDictionary vector description, dictionary matrixAnd with respect toA number of sparse valuesAdjusted sparse vectorDot product of the first orderA number of sparse valuesAn adjusted recording sequenceThe recorded sequence does not containFeatures described by a dictionary vector.
Will be firstA recording sequence and recording methodA first of sparse valuesThe adjusted recording sequence is differenced to obtain a difference valueA first of sparse valuesA sequence of difference valuesThe information contained in the difference sequence is the first information in the recording sequenceThe dictionary vector describes information corresponding to the features. By usingAlgorithm computing a sequence of difference valuesAnd withIs taken as the similarity value ofA first degree of similarity of dictionary vectors, the larger the value, the more indicative ofThe similarity between the information corresponding to the dictionary vector description features and the structural information of the recording sequence is large, so thatThe greater the generalization of the dictionary vectors;
will be firstFirst similarity and second similarity of dictionary vectorThe product of the projection values of the dictionary vectors is taken asGeneralization of dictionary vectorsNormalizing the generalization value of each dictionary vector by a maximum and minimum normalization method, wherein the generalization values of each dictionary vector in the subsequent schemes are normalized normalization values, and the larger the value is, the second order isThe relevance of the information of the feature description corresponding to the dictionary vector and the subject word is larger, and simultaneouslyThe information of the feature description corresponding to each dictionary vector has a greater similarity with the structural information of the recording sequence, so that the probability that the information of the feature description corresponding to the dictionary vector is summarized information is greater. The information described by the dictionary vector is of greater importance.
And then, obtaining the generalization of each dictionary vector according to the correlation condition of each dictionary vector and the topic characteristic vector and the similarity of each dictionary vector and the recording structure information, and carrying out different encryption on each dictionary vector according to the generalization condition of each dictionary vector.
And S003, updating and adjusting the dictionary matrix according to the generalization of each dictionary vector to obtain an updated dictionary matrix, and obtaining ciphertext data according to the updated dictionary matrix.
Because the characteristic information corresponding to the dictionary vector describing the general information of the recording is important for mastering the content of the recording information, the dictionary vector with larger generality needs to be subjected to complex encryption, namely, the data updating amount of the dictionary vector with larger generality is more, and the characteristic described by the dictionary vector with larger generality is strongly changed, so that the general information in the recording information is strongly changed, and the condition that a stolen person masters the recording information due to the leakage of the information is prevented.
Calculating the number of updating data of each dictionary vector:
wherein, the first and the second end of the pipe are connected with each other,denotes the firstThe generalization of the dictionary vector is larger, the larger the value is, the more the characteristic information corresponding to the dictionary vector describes the generalized information in the recording sequence, so that the characteristic information described by the dictionary vector is important for grasping the recording information, thereby ensuring the change amount of the characteristic information of the dictionary vector, effectively ensuring the safety of the encrypted information,the hyperparameter is expressed, M is taken in the scheme,indicating that the symbol is rounded down by a factor,denotes the firstThe number of update data of each dictionary vector.
Performing data update on the dictionary vector according to the number of updated data of the dictionary vector to obtain an updated dictionary vector, as shown in fig. 3:
when the dictionary vectors in the dictionary matrix are updated, in order to avoid updating only the data at fixed positions in each dictionary vector, a partial sequence needs to be acquired in the key sequence to determine the positions of the data to be updated in each dictionary vector, so that the complexity of the positions of the data to be updated can be effectively increased, the original rules of the dictionary vectors can be effectively broken, the cracking difficulty is increased, after the positions of the data to be updated in each dictionary vector are determined, the updated data values also need to be determined, the updated data values of the data to be updated in each dictionary vector need to be determined by utilizing the partial sequence acquired in the key sequence, so that the updating complexity of each dictionary vector can be effectively increased, the original rules of the dictionary matrix can be better broken through the updating method, and the safety of the updated dictionary matrix can be guaranteed. The specific updating method comprises the following steps:
and ordering the dictionary vectors according to the column number of the dictionary vectors in the dictionary matrix from small to large to obtain a dictionary vector sequence.
Obtaining a position sequence and a first key sequence according to the updated data number and the key sequence of each dictionary vector: calculating the cumulative sum of the updated data number of all dictionary vectors as a first number, and intercepting the first position in the key sequenceRecording the sequence at each position as a position sequence, and taking Q as a first number; will be the first in the key sequenceThe data sequence at the location as the first key sequence, as shown in fig. 3;
acquiring the number of elements contained in the dictionary vector, namely the dimension of each dictionary vector, and recording as a first number M; setting the mark position equal to one, and acquiring a dictionary vector at the mark position in the dictionary vector sequence as a target vector, as shown in FIG. 3;
obtaining an updated dictionary vector of the target vector according to the target vector, the position sequence and the first key sequence, as shown in fig. 3, specifically including:
the position before the mark position is obtained is recorded as the position before the markObtaining a first position in the dictionary vector sequenceAll dictionary vectors between the pre-position are recorded to obtain a first dictionary vector set, the accumulated sum of the number of the updated data of all the dictionary vectors in the first dictionary vector set is calculated and recorded as a first accumulated sum, the first accumulated sum is added to obtain a second position, the sum of the first accumulated sum and the number of the updated data of the target vector is recorded as a cut-off position, and the data of the second position obtained in the position sequence is recorded as first dataBased on the first dataThe first key set and the position sequence obtain a middle vector of the target vector, and the method comprises the following steps:
the first dataObtaining the updated position by taking the balance of the first number M, acquiring data at the updated position in the target vector and recording the data as to-be-updated data, acquiring data at the second position in the first key sequence and recording the data as intermediate dataIntermediate data ofThe sum of the target vector and the data to be updated is used as replacement data of the target vector; replacing the replacement data of the target vector with the data to be updated, and keeping other data of the target vector unchanged to obtain a middle vector of the target vector;
and adding one to the second position, repeatedly executing, obtaining a middle vector of the target vector according to the first data, the first key set and the position sequence, and ending when the second position is greater than or equal to the cut-off position. And taking the intermediate vector of the target vector as an updated dictionary vector of the target vector.
And adding one to the mark position, marking the dictionary vector at the mark position in the dictionary vector sequence as a target vector, repeatedly executing, obtaining an updated dictionary vector of the target vector according to the target vector, the position sequence and the first key sequence, ending until the mark position is more than or equal to the number of elements of the dictionary vector, and obtaining an updated dictionary matrix.
Forming the updated dictionary vector into an updated dictionary matrixThus, first of allThe ciphertext data of each recording sequence is:
wherein the content of the first and second substances,is shown asA number of sparse vectors, each of which is a sparse vector,representing the updated dictionary matrix after the update of the dictionary matrix,is shown asCiphertext data of the recording sequence.
And updating the dictionary vectors according to the updated data number and the key sequence to obtain an updated dictionary matrix, and determining the ciphertext data of each recording sequence by using the updated dictionary matrix.
And step S004, decrypting the ciphertext data to obtain a recording sequence.
And the encryption party transmits the ciphertext data of each sound recording sequence and the generalization of each dictionary vector to the decryption party. Because the hyper-parameters of the chaotic mapping function of the key sequence are agreed in advance, the decryption party can determine the chaotic mapping function according to the hyper-parameters, and further obtain the key sequence.
The decryption side takes the ciphertext data of all the recording sequences as input and trains by using a K-SVD algorithm to obtain a dictionary matrixAnd a sparse vector.
And the decryption side obtains the updated data number of each dictionary matrix according to the generalization of each dictionary vector, obtains a position sequence and a first key sequence according to the updated data number and the key sequence of each dictionary vector, and decrypts the ciphertext data by using the reverse process of the encryption method to obtain a recording sequence.
In summary, the embodiments of the present invention provide a voice encryption method for a recording pen, where the summarized information of a recording sequence is helpful for fast mastering the recording sequence information, and therefore, when the recording sequence is encrypted, the summarized information in the recording sequence needs to be encrypted in a complex manner, so as to ensure the security of the summarized information, and when the summarized property of each dictionary vector is analyzed, the generalized property of each dictionary vector is obtained based on the fact that the dictionary vector having the summarized property should be associated with the subject term of the recording sequence, and the generalized property of each dictionary vector also has a property similar to the structural information of the recording sequence, so that each dictionary vector is encrypted differently according to the summarized property of each dictionary vector.
The method comprises the steps of obtaining the number of updating data of each dictionary vector according to the generalization of each dictionary vector, obtaining a position sequence and a first key sequence according to the key sequence and the number of updating data of each dictionary vector, obtaining an updated dictionary matrix according to the number of updating data, the position sequence and the first key sequence of each dictionary vector, and obtaining ciphertext data of each recording sequence according to the updated dictionary matrix and the sparse vector.
It should be noted that: the precedence order of the above embodiments of the present invention is only for description, and does not represent the merits of the embodiments. And that specific embodiments have been described above. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and should not be taken as limiting the scope of the present invention, which is intended to cover any modifications, equivalents, improvements, etc. within the spirit of the present invention.
Claims (6)
1. A voice encryption method for a recording pen is characterized by comprising the following steps:
acquiring a recording sequence to obtain a recording sequence set;
obtaining a dictionary matrix and a plurality of sparse vectors according to the recording sequence set; the dictionary matrix comprises a plurality of dictionary vectors, each element in each sparse vector is called as a sparse value, and one sparse value in each sparse vector corresponds to one dictionary vector;
obtaining a difference sequence of each sparse value according to the relation among the dictionary matrix, the sparse values and the recording sequence; obtaining the theme of each dictionary vector according to the condition that the difference sequence of each sparse value has the theme word characteristics, and taking the dictionary vector with the maximum theme as a theme characteristic vector; taking the projection value of each dictionary vector in the topic feature vector as the projection value of each dictionary vector, and obtaining the first similarity of each dictionary vector according to the similarity of the difference sequence of each sparse value and the structural information of the recording sequence; obtaining the generalization of each dictionary vector according to the first similarity and the projection value of each dictionary vector;
obtaining the updated data number of each dictionary vector according to the generalization of each dictionary vector, obtaining a position sequence and a first key sequence according to the updated data number and the key sequence of each dictionary vector, and obtaining an updated dictionary matrix according to the updated data number, the position sequence and the first key sequence of each dictionary vector; obtaining ciphertext data of each recording sequence according to the updated dictionary matrix and the sparse vector;
the method for obtaining the difference value sequence of each sparse value according to the relation among the dictionary matrix, the sparse values and the recording sequence information comprises the following specific steps:
for any sparse vector, acquiring any one sparse value in the sparse vectors, recording the sparse value as a target sparse value, adjusting the target sparse value to 0, keeping other sparse values unchanged to obtain an adjusted sparse vector of the target sparse value, and obtaining an adjusted recording sequence of the target sparse value according to the dictionary matrix and the adjusted sparse vector of the target sparse value;
calculating the difference value of the recording sequence and the adjusted recording sequence of the target sparse value to obtain a difference value sequence of the target sparse value;
and obtaining a difference value sequence of each sparse value in all the sparse vectors.
2. The voice encryption method for a recording pen according to claim 1, wherein the obtaining of the theme of each dictionary vector according to the condition that the difference sequence of each sparse value has the feature of a subject word comprises the specific steps of:
dividing continuous and uninterrupted non-0 elements in the difference value sequence of each sparse value into a non-zero sequence segment to obtain a plurality of non-zero sequence segments; and calculating the matching value of each non-zero sequence segment and other non-zero sequence segments, taking the mean value of the matching values of each non-zero sequence segment and all other non-zero sequence segments as the comprehensive matching value of each non-zero sequence segment, and taking the mean value of the comprehensive matching values of all non-zero sequence segments as the theme of each dictionary vector.
3. The voice encryption method for recording pens according to claim 1, wherein said obtaining the first similarity of each dictionary vector based on the similarity of the difference sequence of each sparse value and the recording sequence structure information comprises the specific steps of:
and downsampling each recording sequence to obtain a recording sequence after dimensionality reduction, and calculating the similarity between the difference sequence of each sparse value and the recording sequence after dimensionality reduction to obtain the first similarity of each dictionary vector.
4. The voice encryption method for the recording pen according to claim 1, wherein the obtaining of the updated dictionary matrix according to the updated data number, the position sequence and the first key sequence of each dictionary vector comprises the specific steps of:
setting the mark position to one; recording the dictionary vector of the marked position in the dictionary vector sequence as a target vector, recording the number of elements contained in the target vector as a first number, and obtaining an updated dictionary vector of the target vector according to the target vector, the position sequence and the first key sequence, wherein the method comprises the following steps:
acquiring dictionary vectors between a first position in the dictionary vector sequence and a position before the target vector to obtain a dictionary vector set, and obtaining a plurality of second positions according to the number of updated data of each dictionary vector in the dictionary vector set; acquiring data at a plurality of second positions in the position sequence to obtain a plurality of first data; taking the first data and the remainder of the first number as an updating position to obtain a plurality of updating positions; acquiring data at a plurality of updating positions in a target vector to obtain a plurality of data to be updated, acquiring data at a plurality of second positions in a first key sequence to obtain a plurality of intermediate data, acquiring a plurality of replacement data according to the plurality of intermediate data and the plurality of data to be updated, and replacing the plurality of data to be updated in the target vector by using the corresponding replacement data respectively to obtain an updated dictionary vector of the target vector;
adding one to the mark position, marking the dictionary vector at the mark position in the dictionary vector sequence as a target vector, and repeatedly executing: and obtaining an updated dictionary vector of the target vector according to the target vector, the position sequence and the first key sequence, and ending when the mark position is more than or equal to the number of elements of the dictionary vector to obtain an updated dictionary matrix.
5. The method for encrypting the voice of the recording pen according to claim 1, wherein the obtaining of the ciphertext data of each recording sequence according to the updated dictionary matrix and the sparse vector comprises the specific steps of:
6. The voice encryption method for the recording pen according to claim 1, wherein the obtaining of the position sequence and the first key sequence according to the updated data number and the key sequence of each dictionary vector comprises the specific steps of:
calculating the sum of the updated data numbers of all dictionary vectors to obtain a first number, and intercepting the first position in the key sequence toThe sequence at a position is denoted as the sequence of positions,taking a first number; the position sequence is cut out from the key sequence to obtain a first key sequence.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211629107.3A CN115618397B (en) | 2022-12-19 | 2022-12-19 | Voice encryption method for recording pen |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211629107.3A CN115618397B (en) | 2022-12-19 | 2022-12-19 | Voice encryption method for recording pen |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115618397A CN115618397A (en) | 2023-01-17 |
CN115618397B true CN115618397B (en) | 2023-03-03 |
Family
ID=84880005
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211629107.3A Active CN115618397B (en) | 2022-12-19 | 2022-12-19 | Voice encryption method for recording pen |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115618397B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116308813B (en) * | 2023-05-17 | 2023-08-08 | 青岛农村商业银行股份有限公司 | Different-industry combined financial equity safety management system |
CN116366373B (en) * | 2023-06-01 | 2023-08-22 | 深圳市柏英特电子科技有限公司 | Intelligent management method and storage medium for set top box data |
CN117131530B (en) * | 2023-10-20 | 2024-01-30 | 合肥亚明汽车部件有限公司 | Intelligent factory sensitive data encryption protection method |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102034478B (en) * | 2010-11-17 | 2013-10-30 | 南京邮电大学 | Voice secret communication system design method based on compressive sensing and information hiding |
US8484024B2 (en) * | 2011-02-24 | 2013-07-09 | Nuance Communications, Inc. | Phonetic features for speech recognition |
EP3166020A1 (en) * | 2015-11-06 | 2017-05-10 | Thomson Licensing | Method and apparatus for image classification based on dictionary learning |
CN106056526B (en) * | 2016-05-26 | 2019-04-12 | 南昌大学 | A kind of resume image based on parsing rarefaction representation and compressed sensing |
CA3097655A1 (en) * | 2019-10-30 | 2021-04-30 | Royal Bank Of Canada | System and method for machine learning architecture with differential privacy |
CN114556328A (en) * | 2019-12-31 | 2022-05-27 | 深圳市欢太科技有限公司 | Data processing method and device, electronic equipment and storage medium |
-
2022
- 2022-12-19 CN CN202211629107.3A patent/CN115618397B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN115618397A (en) | 2023-01-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115618397B (en) | Voice encryption method for recording pen | |
Niu et al. | A hybrid R-BILSTM-C neural network based text steganalysis | |
Husain et al. | REMAP: Multi-layer entropy-guided pooling of dense CNN features for image retrieval | |
US20200184016A1 (en) | Segment vectors | |
US11582237B2 (en) | Systems and methods for privacy-protecting hybrid cloud and premise stream processing | |
WO2011052412A1 (en) | Speech recognition system, speech recognition request device, speech recognition method, speech recognition program, and recording medium | |
CN113051371B (en) | Chinese machine reading understanding method and device, electronic equipment and storage medium | |
WO2020011200A1 (en) | Cross-domain data fusion method and system, and storage medium | |
JP7140317B2 (en) | Method for learning data embedding network that generates marked data by synthesizing original data and mark data, method for testing, and learning device using the same | |
Chee et al. | Cancellable speech template via random binary orthogonal matrices projection hashing | |
Li et al. | Topic-aware neural linguistic steganography based on knowledge graphs | |
CN111966811A (en) | Intention recognition and slot filling method and device, readable storage medium and terminal equipment | |
CN113705727B (en) | Decision tree modeling method, prediction method, equipment and medium based on differential privacy | |
CN116881739B (en) | Ciphertext security retrieval method oriented to similarity of spatial keywords | |
US11093474B2 (en) | Computer architecture for emulating multi-dimensional string correlithm object dynamic time warping in a correlithm object processing system | |
CN113239852B (en) | Privacy image processing method, device and equipment based on privacy protection | |
CN115048661A (en) | Model processing method, device and equipment | |
CN114283811A (en) | Voice conversion method, device, computer equipment and storage medium | |
Satheesh Kumar et al. | Content-Based Video Retrieval Based on Security Using Enhanced Video Retrieval System with Region-Based Neural Network (EVRS-RNN) and K-Means Classification | |
JPWO2004088591A1 (en) | Verification device and registration device | |
JP4746004B2 (en) | Registration device | |
CN117272353B (en) | Data encryption storage protection system and method | |
US11954676B1 (en) | Apparatus and method for minting NFTs from user-specific events | |
AU2022215149B2 (en) | Synthesizing user transactional data for de- identifying sensitive information | |
US10997143B2 (en) | Computer architecture for emulating single dimensional string correlithm object dynamic time warping in a correlithm object processing system |
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 |