An intelligent authentication method and system based on voiceprint recognition

Abstract

The present invention discloses an intelligent authentication method and system based on voiceprint recognition, which relates to the technical field of voiceprint recognition, including collecting user voice signals for preprocessing, extracting voiceprint features of the preprocessed voice signals; performing feature transformation on the extracted voiceprint features and generating voiceprint templates, and re-collecting user voice signals for authentication verification; displaying user authentication results in real time through a visual interface, and storing the sound data generated during the collection and analysis process. The present invention collects user voice signals for preprocessing, extracts voiceprint features of the preprocessed voice signals, performs feature transformation on the extracted voiceprint features and generates voiceprint templates, and re-collects user voice signals for authentication verification, which greatly improves the quality of voiceprint data and the robustness of the recognition system, enhances the anti-interference ability and security of the system, improves the recognition accuracy, and ensures the security and processing efficiency of the data.

Description

Technical Field

[0001] The present invention relates to the technical field of voiceprint recognition, and in particular to an intelligent authentication method and system based on voiceprint recognition. Background Art

[0002] In today's society, voiceprint recognition technology, as an emerging means of biometrics, has gradually become a hot spot for research and application. Compared with biometrics such as fingerprints and irises, voiceprint recognition has the advantages of simple operation, strong non-invasiveness, and difficulty in forging. The uniqueness of voiceprints stems from the differences in each person's vocal organs and vocal habits, which makes voiceprints an effective way of identity authentication. With the development of artificial intelligence technology, especially the application of deep learning, the accuracy and efficiency of voiceprint recognition have been significantly improved. Various algorithms such as convolutional neural networks (CNN) and recurrent neural networks (RNN) are widely used in the extraction and recognition of voiceprint features, which significantly optimizes the learning ability and generalization of the model.

[0003] Although the existing technology has made certain progress, it still faces many challenges in practical applications. The interference of background noise is a difficult problem to overcome in voiceprint recognition. Most existing methods rely on fixed feature extraction and matching algorithms, which cannot fully utilize the diversity and complexity of user voices. The existing voiceprint template generation and storage methods are also insufficient in data security and privacy protection, and are vulnerable to attacks and tampering. Summary of the invention

[0004] In view of the problems existing in the above-mentioned existing intelligent authentication methods and systems based on voiceprint recognition, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is that most of the existing methods rely on fixed feature extraction and matching algorithms, which cannot fully utilize the diversity and complexity of user voices. The existing voiceprint template generation and storage methods also have deficiencies in data security and privacy protection, and are vulnerable to attacks and tampering.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: an intelligent authentication method based on voiceprint recognition, which includes collecting user voice signals for preprocessing, extracting voiceprint features of the preprocessed voice signals; performing feature transformation on the extracted voiceprint features and generating a voiceprint template, and re-collecting user voice signals for authentication; displaying the user authentication results in real time through a visual interface, and storing the sound data generated during the collection and analysis process.

[0007] As a preferred solution of the intelligent authentication method based on voiceprint recognition of the present invention, wherein: the collecting of user voice signals for preprocessing refers to using a microphone to collect user voice signals;

[0008] Use noise reduction algorithms to filter out background noise from sound signals;

[0009] Normalize the amplitude of the sound signal to the range of [-1,1];

[0010] The normalized sound signal is divided into several frames, each frame is 25ms long and the frame shift is 10ms.

[0011] As a preferred solution of the intelligent authentication method based on voiceprint recognition of the present invention, wherein: the extraction of voiceprint features of the pre-processed sound signal refers to performing a short-time Fourier transform STFT on each frame of the sound signal to obtain a spectrum diagram;

[0012] Use the Mel filter to map the spectrum to the Mel frequency to obtain the Mel spectrum;

[0013] Perform logarithmic transformation on the Mel spectrum;

[0014] The logarithmically transformed Mel spectrum is subjected to discrete cosine transform DCT to obtain MFCC features.

[0015] As a preferred solution of the intelligent authentication method based on voiceprint recognition of the present invention, wherein: the feature transformation of the extracted voiceprint features and the generation of a voiceprint template refers to randomly generating an orthogonal matrix based on MFCC features using the Gram-Schmidt orthogonalization process, and defining the size of the orthogonal matrix as d×d;

[0016] The MFCC features are multiplied by the orthogonal matrix to perform random projection changes to obtain high-dimensional features;

[0017] Use sparse coding algorithm to sparsely encode high-dimensional features, obtain sparse coding feature vectors and construct sparse coding feature matrix Z;

[0018] Randomly extract samples from the sparse coding feature matrix Z as a training set;

[0019] Randomly select K samples from the training set as the initial codebook vector, the formula is:

[0020]

[0021] Among them C (0) is the initial code book, is the Kth codebook vector;

[0022] Assign the sparse coding feature vector with the smallest distance to each codebook vector. The formula is:

[0023]

[0024] in is the vector assigned to the Kth codebook in the tth iteration The set of sparse coded feature vectors, is the jth codebook vector in the tth iteration, is the Euclidean distance between the sparse coding feature vector and the Kth codebook vector, is the minimum distance between the sparse coding feature vector and the codebook vector in the tth iteration, z is the sparse coding feature vector;

[0025] Based on each codebook vector, each codebook vector is updated using the arithmetic mean of all sparse coded feature vectors in the assigned sparse coded feature vector set, as follows:

[0026]

[0027] in is the Kth codebook vector after update;

[0028] A threshold is set based on historical data, and the iteration operation is continued until the updated code book vector value is less than the predetermined threshold, and then the iteration is stopped;

[0029] Based on the number of codebook vectors K, an equal number of false point vectors are generated, and the formula is:

[0030] P K =μ K +σ K ∈ K ,

[0031] Where P K is the generated false point vector, μ K and σ K are the mean and standard deviation calculated based on the Kth codebook vector, ∈ K are random numbers generated from a standard normal distribution;

[0032] The false point vector is mixed with the code book vector to form the voiceprint template C', the formula is:

[0033]

[0034] Where C' is the voiceprint template containing the code book vector and the false point vector, is the updated codebook vector;

[0035] Generate a binary index I of length L for the codebook vector in the voiceprint template i ;

[0036] Use the binary OR operation to aggregate the binary index of the code book vector in the voiceprint template to generate the global key A. The formula is:

[0037]

[0038] Where N is the total number of codebook vectors in the voiceprint template.

[0039] As a preferred solution of the intelligent authentication method based on voiceprint recognition of the present invention, wherein: the re-collection of the user's voice signal for authentication verification refers to re-collecting the user's current voice data x new , preprocess the collected data and extract the current MFCC features;

[0040] Load the voiceprint template C' and global key A corresponding to the user;

[0041] Use random projection and sparse coding to transform the current MFCC features to obtain the current test feature vector Y u ;

[0042] Calculate the Euclidean distance between the current test feature vector and the codebook vector in the voiceprint template

[0043] Generate a binary index and a global key from the current test feature vector, compare it with the binary index and global key of the voiceprint template C', and select the voiceprint template code book vector that matches the current test feature vector;

[0044] Setting the distortion metric threshold T based on the selected voiceprint template codebook vector th ;

[0045] The average distortion metric T is calculated for the matched current test feature vector, and the formula is:

[0046]

[0047] Where T' is the number of the screened voiceprint template codebook vectors, and u is the index of the current test feature vector;

[0048] The average distortion metric T is compared with the preset distortion metric threshold T th The result of user identity authentication is obtained by comparison. If the average distortion metric T ≤ the preset distortion metric threshold T th , then the match is successful and the user identity is determined. If the average distortion metric T> the preset distortion metric threshold T th , the matching fails and the user identity verification fails.

[0049] As a preferred solution of the intelligent authentication method based on voiceprint recognition described in the present invention, wherein: the real-time display of the user authentication result through a visual interface refers to using the front-end technology stack Vue combined with the Bootstrap framework to build a visual interface, including a data collection area, a data processing display area and a verification result area;

[0050] Display the collected user voice data and user identity data in real time on the visual interface;

[0051] Based on the results of user authentication, successfully matched users are allowed to access, and unmatched users are provided with a re-verification option, allowing users to try voice input for authentication again.

[0052] As a preferred solution of the intelligent authentication method based on voiceprint recognition described in the present invention, wherein: the storage of sound data generated during the collection and analysis process refers to storing the collected sound data, the voiceprint data generated by the analysis and the user authentication data in a database, the database classifies and stores the data according to the timestamp data and implements encryption protection measures, synchronously uploads the stored data to the cloud for backup, and regularly performs integrity checks on the stored data and backup data.

[0053] Another object of the present invention is to provide an intelligent authentication method system based on voiceprint recognition, which comprises:

[0054] The collection and extraction module is used to collect the user's voice signal for preprocessing and extract the voiceprint features of the preprocessed voice signal;

[0055] The template recognition module is used to transform the extracted voiceprint features and generate a voiceprint template, and re-collect the user's voice signal for identity verification;

[0056] The visual storage module is used to display the user authentication results in real time through a visual interface and store the sound data generated during the collection and analysis process.

[0057] A computer device comprises: a memory and a processor; the memory stores a computer program, and the processor implements the steps of the above-mentioned intelligent authentication method based on voiceprint recognition when executing the computer program.

[0058] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned intelligent authentication method based on voiceprint recognition.

[0059] The beneficial effects of the present invention are as follows: the present invention collects user voice signals for preprocessing, extracts voiceprint features of the preprocessed voice signals, performs feature transformation on the extracted voiceprint features and generates voiceprint templates, and re-collects user voice signals for identity verification, thereby greatly improving the quality of voiceprint data and the robustness of the recognition system, enhancing the system's anti-interference ability and security, and improving recognition accuracy while also ensuring data security and processing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0061] Figure 1 The figure is a flowchart of the intelligent authentication method based on voiceprint recognition.

[0062] Figure 2 This is a schematic diagram of the structure of the intelligent identity verification system based on voiceprint recognition.

[0063] Figure 3 The figure is a schematic diagram of the structure of the intelligent authentication method based on voiceprint recognition. DETAILED DESCRIPTION

[0064] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0065] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0066] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0067] Example 1, reference Figure 1 and Figure 3 , which is the first embodiment of the present invention, and provides an intelligent authentication method based on voiceprint recognition. The intelligent authentication method based on voiceprint recognition includes:

[0068] S1. Collect user voice signals for preprocessing and extract voiceprint features of the preprocessed voice signals;

[0069] Specifically, collecting the user voice signal for preprocessing means using a microphone to collect the user voice signal;

[0070] Use noise reduction algorithms to filter out background noise from sound signals;

[0071] Normalize the amplitude of the sound signal to the range of [-1,1];

[0072] The normalized sound signal is divided into several frames, each frame is 25ms long and the frame shift is 10ms.

[0073] The use of microphones and noise reduction algorithms can significantly improve the quality of sound signals, thereby increasing the accuracy of the voiceprint recognition system. Under changing environmental conditions, through effective noise reduction and normalization processing, the system can resist the impact of background noise and device differences and maintain stable recognition performance. This is especially important for mobile devices and public safety fields. By properly framing the sound signal, feature extraction and subsequent processing can be made more efficient. The stability of short-term signals allows the use of simpler models for rapid feature analysis, thereby reducing computing costs and improving the response speed of the system.

[0074] Furthermore, extracting the voiceprint features of the pre-processed sound signal refers to performing a short-time Fourier transform (STFT) on each frame of the sound signal. By performing STFT processing on each frame of 25 milliseconds of sound signal, a time-frequency representation, i.e., a spectrogram, can be obtained;

[0075] Use the Mel filter to map the spectrogram to the Mel frequency. The Mel frequency scale is more consistent with the perceptual characteristics of the human auditory system, making the extracted voiceprint features closer to the auditory experience of the human ear, and obtaining the Mel spectrogram;

[0076] Perform logarithmic transformation on the Mel spectrum;

[0077] The logarithmically transformed Mel spectrum is subjected to discrete cosine transform DCT to obtain MFCC features.

[0078] Combining the Mel filter and logarithmic transform, the auditory characteristics of the human ear can be better simulated, and features that are more in line with the perceptual characteristics can be extracted, thereby improving the accuracy of voiceprint recognition. The logarithmic transform can effectively compress the dynamic range of the spectrum, reduce the impact of frequency components with large amplitudes, and improve the robustness of the features. The discrete cosine transform further extracts the inverse spectrum features of the speech signal, making the features more robust in different environments and able to cope with variable background noise and different recording equipment. The low-dimensional features not only reduce the amount of data processing computation, but also retain the main information of the speech signal, which helps to improve the processing efficiency of the voiceprint recognition system.

[0079] S2, transform the extracted voiceprint features and generate a voiceprint template, and re-collect the user's voice signal for identity verification;

[0080] Specifically, performing feature transformation on the extracted voiceprint features and generating a voiceprint template refers to randomly generating an orthogonal matrix based on the MFCC features using a Gram-Schmidt orthogonalization process, and defining the size of the orthogonal matrix to be d×d, where d is the dimension of the MFCC coefficients;

[0081] The MFCC features are multiplied by the orthogonal matrix to perform random projection changes to obtain high-dimensional features;

[0082] Use sparse coding algorithm to sparsely encode high-dimensional features, obtain sparse coding feature vectors and construct sparse coding feature matrix Z;

[0083] Randomly extract samples from the sparse coding feature matrix Z as a training set;

[0084] Randomly select K samples from the training set as the initial codebook vector, the formula is:

[0085]

[0086] Among them C (0) is the initial code book, is the Kth codebook vector;

[0087] Assign the sparse coding feature vector with the smallest distance to each codebook vector. Our goal is to divide the entire data set into K different categories. Each codebook vector represents a category, and each category is composed of all the feature vectors with the closest distance to it. The formula is:

[0088]

[0089] in is the vector assigned to the Kth codebook in the tth iteration The set of sparse coded feature vectors, is the jth codebook vector in the tth iteration, is the Euclidean distance between the sparse coding feature vector and the Kth codebook vector, is the minimum distance between the sparse coding feature vector and the codebook vector in the tth iteration, z is the sparse coding feature vector;

[0090] Based on each codebook vector, each codebook vector is updated using the arithmetic mean of all sparse coded feature vectors in the assigned sparse coded feature vector set, as follows:

[0091]

[0092] in is the Kth codebook vector after update;

[0093] A threshold is set based on historical data, and the iteration operation is continued until the updated code book vector value is less than the predetermined threshold, and then the iteration is stopped;

[0094] Based on the number of codebook vectors K, an equal number of false point vectors are generated, and the formula is:

[0095] P K =μ K +σ K ∈ K ,

[0096] Where P K is the generated false point vector, μ K and σ K are the mean and standard deviation calculated based on the Kth codebook vector, ∈ K are random numbers generated from a standard normal distribution;

[0097] The false point vector is mixed with the code book vector to form the voiceprint template C', the formula is:

[0098]

[0099] Where C' is the voiceprint template containing the code book vector and the false point vector, is the updated codebook vector;

[0100] Generate a binary index I of length L for the codebook vector in the voiceprint template i ;

[0101] Use the binary OR operation to aggregate the binary index of the code book vector in the voiceprint template to generate the global key A. The formula is:

[0102]

[0103] Where N is the total number of codebook vectors in the voiceprint template.

[0104] High-dimensional features have richer information and better distinguishing ability, which significantly improves the accuracy of voiceprint recognition, especially when dealing with complex voice signals and changing environments. This method can effectively reduce the false recognition rate. The sparse coding algorithm extracts the main information in the data by sparsely encoding high-dimensional features, reducing the interference of noise and redundant information. This process makes the generated features more robust and can maintain high recognition performance in different environments and devices. Through sparse coding and codebook vector generation, a large number of high-dimensional features can be compressed into a small number of codebook vectors and false point vectors. This compression method not only greatly reduces the space requirements for data storage, but also improves the processing efficiency of subsequent feature matching and identity authentication, enabling the system to respond quickly to user requests. By generating false point vectors and mixing them with codebook vectors, a voiceprint template containing real codebook vectors and false point vectors is formed, which effectively increases the security of the data. It is difficult for attackers to distinguish between real features and false features, further protecting the privacy of users. The global key A is generated and the voiceprint template is aggregated by binary index, which enhances the security of data during transmission and storage.

[0105] Further, re-collecting the user's voice signal for identity verification refers to re-collecting the user's current voice data x new , preprocess the collected data and extract the current MFCC features;

[0106] Load the voiceprint template C' and global key A corresponding to the user;

[0107] Use random projection and sparse coding to transform the current MFCC features to obtain the current test feature vector Y u ;

[0108] Calculate the Euclidean distance between the current test feature vector and the codebook vector in the voiceprint template

[0109] Generate a binary index and a global key from the current test feature vector, compare it with the binary index and global key of the voiceprint template C', and select the voiceprint template code book vector that matches the current test feature vector;

[0110] Setting the distortion metric threshold T based on the selected voiceprint template codebook vector th ;

[0111] The average distortion metric T is calculated for the matched current test feature vector, and the formula is:

[0112]

[0113] Where T' is the number of the screened voiceprint template codebook vectors, and u is the index of the current test feature vector;

[0114] The average distortion metric T is compared with the preset distortion metric threshold T th The result of user identity authentication is obtained by comparison. If the average distortion metric T ≤ the preset distortion metric threshold T th , then the match is successful and the user identity is determined. If the average distortion metric T> the preset distortion metric threshold T th , the matching fails and the user identity verification fails.

[0115] By re-collecting the user's current voice data and extracting MFCC features, the present invention can capture the latest voiceprint information. Combined with random projection and sparse coding technology, the generated high-dimensional feature vector is more discriminative and robust, effectively improving the accuracy of identity verification. Re-collecting the user's voice data and pre-processing it can remove background noise and other interference factors, making the extracted features purer and more stable. Through binary indexing and global keys, the present invention realizes a fast feature matching and verification process. The use of binary indexing greatly reduces the computational complexity and improves the response speed of the system. The application of global keys enhances data security and prevents malicious attacks and tampering. The global key is generated by aggregating the index of the code book vector in the voiceprint template, effectively increasing the security of the data. It is difficult for attackers to obtain complete voiceprint template information through a single feature vector, thereby protecting the privacy of users.

[0116] S3, display the user authentication results in real time through a visual interface, and store the sound data generated during the collection and analysis process;

[0117] Specifically, displaying the user verification results in real time through a visual interface means using the front-end technology stack Vue combined with the Bootstrap framework to build a visual interface, including a data collection area, a data processing display area, and a verification result area;

[0118] Display the collected user voice data and user identity data in real time on the visual interface;

[0119] Based on the results of user authentication, successfully matched users are allowed to access, and unmatched users are provided with a re-verification option, allowing users to try voice input for authentication again.

[0120] By using the visual interface built with Vue and Bootstrap, users can intuitively see the processing process of their voice data and the results of identity verification. This transparent display method enhances the user's sense of trust and satisfaction, making the entire identity verification process more humane and friendly. Based on the results of user identity verification, the visual interface allows successfully matched users to access and provides re-verification options for unmatched users. This design simplifies the operation process and enables users to complete all verification operations on one interface without frequently switching pages or applications.

[0121] Furthermore, storing the sound data generated during the collection and analysis process means storing the collected sound data, the voiceprint data generated by the analysis, and the user identity data in a database. The database classifies and stores the data according to timestamp data and implements encryption protection measures, simultaneously uploading the stored data to the cloud for backup, and regularly performing integrity checks on the stored data and backup data.

[0122] By storing the collected sound data, the voiceprint data generated by analysis, and the user authentication data in the database and implementing encryption protection measures, data leakage and unauthorized access can be effectively prevented. Encryption protection can ensure that even if the data is intercepted, unauthorized users cannot read and use the data, thereby improving the security of the system. Using timestamps to classify and store data can make data management and retrieval more efficient. Through timestamps, data within a specific time period can be quickly located, data access speed can be improved, and system performance can be optimized. This measure is particularly important for large-scale data storage and rapid retrieval. Regular integrity testing of stored data and backup data can promptly detect and correct data errors to ensure data integrity and consistency. Cloud backup provides high reliability and high availability, ensuring that data can be quickly restored when local equipment fails or data is damaged, ensuring system continuity and reliability.

[0123] Example 2, reference Figure 2 This is the second embodiment of the present invention. This embodiment is different from the previous embodiment and provides an intelligent authentication method system based on voiceprint recognition, including a collection and extraction module for collecting user voice signals for preprocessing and extracting voiceprint features of the preprocessed voice signals;

[0124] The template recognition module is used to transform the extracted voiceprint features and generate a voiceprint template, and re-collect the user's voice signal for identity verification;

[0125] The visual storage module is used to display the user authentication results in real time through a visual interface and store the sound data generated during the collection and analysis process.

[0126] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0127] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0128] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0129] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

Claims

1. An intelligent authentication method based on voiceprint recognition, characterized by: include, Collect user voice signals for preprocessing, and extract voiceprint features of the preprocessed voice signals; The extracted voiceprint features are transformed and a voiceprint template is generated, and the user's voice signal is collected again for identity verification; Display user authentication results in real time through a visual interface, and store the sound data generated during the collection and analysis process; Collecting the user's voice signal for preprocessing refers to using a microphone to collect the user's voice signal; Use noise reduction algorithms to filter out background noise from sound signals; Normalize the amplitude of the sound signal to the range of [-1,1]; The normalized sound signal is divided into several frames, each frame is 25ms long and the frame shift is 10ms; Extracting the voiceprint features of the pre-processed sound signal refers to performing a short-time Fourier transform (STFT) on each frame of the sound signal to obtain a spectrum diagram; Use the Mel filter to map the spectrum to the Mel frequency to obtain the Mel spectrum; Perform logarithmic transformation on the Mel spectrum; Perform discrete cosine transform (DCT) on the logarithmic transformed Mel spectrum to obtain MFCC features; The feature transformation of the extracted voiceprint features and generating a voiceprint template refers to randomly generating an orthogonal matrix based on the MFCC features using the Gram-Schmidt orthogonalization process, defining the size of the orthogonal matrix as d×d; performing random projection changes on the multiplication of the MFCC features and the orthogonal matrix to obtain high-dimensional features; Use sparse coding algorithm to sparsely encode high-dimensional features, obtain sparse coding feature vectors and construct sparse coding feature matrix Z; Randomly extract samples from the sparse coding feature matrix Z as a training set; Randomly select K samples from the training set as the initial codebook vector, the formula is: Among them C (0) is the initial code book, is the Kth codebook vector; Assign the sparse coding feature vector with the smallest distance to each codebook vector. The formula is: in is the vector assigned to the Kth codebook in the tth iteration The set of sparse coded feature vectors, is the jth codebook vector in the tth iteration, is the Euclidean distance between the sparse coding feature vector and the Kth codebook vector, is the minimum distance between the sparse coding feature vector and the codebook vector in the tth iteration, z is the sparse coding feature vector; Based on each codebook vector, each codebook vector is updated using the arithmetic mean of all sparse coded feature vectors in the assigned sparse coded feature vector set, as follows: in is the Kth codebook vector after update; A threshold is set based on historical data, and the iteration operation is continued until the updated code book vector value is less than the predetermined threshold, and then the iteration is stopped; Based on the number of codebook vectors K, an equal number of false point vectors are generated, and the formula is: P K =μ K +s K ·∈ K , Where P K is the generated false point vector, μ K and σ K are the mean and standard deviation calculated based on the Kth codebook vector, ∈ K are random numbers generated from a standard normal distribution; The false point vector is mixed with the code book vector to form the voiceprint template C′, the formula is: Where C′ is the voiceprint template containing the codebook vector and the false point vector, is the updated codebook vector; Generate a binary index I of length L for the codebook vector in the voiceprint template i ; Use the binary OR operation to aggregate the binary index of the code book vector in the voiceprint template to generate the global key A. The formula is: Where N is the total number of codebook vectors in the voiceprint template.

2. The intelligent authentication method based on voiceprint recognition as claimed in claim 1, characterized in that: The recollection of the user's voice signal for identity verification refers to recollecting the user's current voice data x new , preprocess the collected data and extract the current MFCC features; Load the voiceprint template c′ and global key A corresponding to the user; Use random projection and sparse coding to transform the current MFCC features to obtain the current test feature vector Y u ; Calculate the Euclidean distance between the current test feature vector and the codebook vector in the voiceprint template Generate a binary index and a global key from the current test feature vector, compare it with the binary index and global key of the voiceprint template C', and select the voiceprint template code book vector that matches the current test feature vector; Setting the distortion metric threshold T based on the selected voiceprint template codebook vector th ; The average distortion metric T is calculated for the matched current test feature vector, and the formula is: Where T′ is the number of the screened voiceprint template codebook vectors, and u is the index of the current test feature vector; The average distortion metric T is compared with the preset distortion metric threshold T th The result of user identity authentication is obtained by comparison. If the average distortion metric T ≤ the preset distortion metric threshold T th , then the match is successful and the user identity is determined. If the average distortion metric T> the preset distortion metric threshold T th , the matching fails and the user identity verification fails.

3. The intelligent authentication method based on voiceprint recognition as claimed in claim 2, characterized in that: The real-time display of user authentication results through a visual interface refers to using the front-end technology stack Vue combined with the Bootstrap framework to build a visual interface, including a data collection area, a data processing display area, and a verification result area; Display the collected user voice data and user identity data in real time on the visual interface; Based on the results of user authentication, successfully matched users are allowed to access, and unmatched users are provided with a re-verification option, allowing users to try voice input for authentication again.

4. The intelligent authentication method based on voiceprint recognition as claimed in claim 3, characterized in that: The storage of sound data generated during the collection and analysis process refers to storing the collected sound data, the voiceprint data generated by the analysis, and the user identity data in a database. The database classifies and stores the data according to timestamp data and implements encryption protection measures, simultaneously uploads the stored data to the cloud for backup, and regularly performs integrity checks on the stored data and backup data.

5. An intelligent authentication system based on voiceprint recognition based on the intelligent authentication method based on voiceprint recognition according to any one of claims 1 to 4, characterized in that: include, The collection and extraction module is used to collect the user's voice signal for preprocessing and extract the voiceprint features of the preprocessed voice signal; The template recognition module is used to transform the extracted voiceprint features and generate a voiceprint template, and re-collect the user's voice signal for identity verification; The visual storage module is used to display the user authentication results in real time through a visual interface and store the sound data generated during the collection and analysis process.

6. A computer device comprising: Memory and processor; The memory stores a computer program, characterized in that: when the processor executes the computer program, the steps of the intelligent authentication method based on voiceprint recognition described in any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the intelligent authentication method based on voiceprint recognition described in any one of claims 1 to 4 are implemented.

Images