Large-scale voiceprint database management method and device based on cloud service

Abstract

The invention provides a large-scale voiceprint database management method and device based on cloud service, and relates to the technical field of computers. The method comprises the following steps: receiving and storing voiceprint data of a user to a cloud distributed storage system, preprocessing the voiceprint data, extracting a corresponding voiceprint feature vector, and storing the voiceprint feature vector in a distributed index structure; obtaining corresponding voiceprint data to be verified, performing preprocessing based on the voiceprint data to be verified to obtain a corresponding voiceprint feature vector to be verified, and determining a corresponding verification feature vector based on a stored voiceprint feature vector most similar to the voiceprint feature vector to be verified, which is quickly retrieved in the distributed index structure; and calculating a similarity score corresponding to the to-be-verified voiceprint feature vector and the verification feature vector, and determining a corresponding identity verification result according to the similarity score. The voiceprint database management method and device solve the problem that the overall performance and safety of large-scale voiceprint database management are not high in the prior art.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a large-scale voiceprint database management method and device based on cloud services. Background Art

[0002] With the rapid development of biometric technology, voiceprint recognition, as a contactless means of identity authentication, has been widely used in many fields such as finance, security, and medical care. Traditional voiceprint database management systems usually rely on local servers for data storage and processing, which will encounter problems such as limited storage capacity, insufficient computing power, and poor scalability when facing large-scale user groups.

[0003] Specifically, when the number of users increases dramatically, a single server cannot provide enough storage space to accommodate the rapidly growing voiceprint data; at the same time, the complex voiceprint matching algorithm has high demand for computing resources, which makes the system response slow under high concurrent requests, affecting the user experience. In addition, the centralized storage model also has a high security risk, and once attacked, it may lead to the leakage of a large amount of sensitive information.

[0004] From the above, we can see that how to improve the overall performance and security of large-scale voiceprint database management remains to be solved. Summary of the invention

[0005] The present application provides a large-scale voiceprint database management method and device based on cloud services, which can solve the problem of low overall performance and security of large-scale voiceprint database management in related technologies. The technical solution is as follows:

[0006] According to one aspect of the present application, a large-scale voiceprint database management method based on cloud services includes: receiving and storing the user's voiceprint data to a cloud-based distributed storage system, preprocessing the voiceprint data, extracting the corresponding voiceprint feature vector, and storing the voiceprint feature vector in a distributed index structure; when an identity authentication request is received, obtaining the corresponding voiceprint data to be verified, preprocessing the voiceprint data to be verified to obtain the corresponding voiceprint feature vector to be verified, and determining the corresponding verification feature vector based on the quick retrieval of the stored voiceprint feature vector that is most similar to the voiceprint feature vector to be verified in the distributed index structure; calculating the similarity score between the voiceprint feature vector to be verified and the verification feature vector, and determining the corresponding identity authentication result according to the similarity score.

[0007] In an exemplary embodiment, the method also includes: receiving a voiceprint data update request submitted by a user through a client application, and sending guidance information to the client based on the voiceprint data update request to guide the user to record a new voiceprint sample; obtaining the latest voiceprint sample recorded by the user, preprocessing the latest voiceprint sample, and extracting the latest voiceprint sample feature vector; storing the latest voiceprint sample feature vector in a cloud-based distributed storage system, updating the distributed index structure, and sending a notification of successful update to the user to require the user to perform identity authentication to confirm that the update is successful.

[0008] In an exemplary embodiment, voiceprint data is collected at each identity authentication or periodically, and the method further includes: obtaining a voiceprint sample of the user and retrieving an existing voiceprint model; comparing the voiceprint sample with the existing voiceprint model to determine whether a change has occurred; if a change is detected, updating the existing voiceprint model based on an incremental learning algorithm and the voiceprint sample.

[0009] In an exemplary embodiment, the method also includes: deploying data centers in different geographical areas, determining corresponding primary data centers and backup data centers, wherein the voiceprint data between each of the data centers remains consistent; monitoring the status of the primary data center and all the backup data centers, and if an abnormality is detected, issuing an alarm and initiating corresponding recovery measures; when the primary data center fails, switching to the corresponding backup data center.

[0010] In an exemplary embodiment, the method also includes: obtaining an indicator data set corresponding to the database management system, the indicator data set including CPU usage, memory occupancy, network bandwidth and service request volume; detecting the indicator data set, and if an increase in load is detected, adding computing resources to process additional requests; if the load decreases, determining the corresponding excess resources; when a failure is detected in a node of the database management system, determining the corresponding faulty node, and transferring the tasks corresponding to the faulty node to other healthy nodes.

[0011] In an exemplary embodiment, in the process of determining the corresponding identity authentication result based on the similarity score, the method also includes: calling a preset similarity verification threshold; comparing the similarity score with the similarity verification threshold; if the similarity score is greater than the similarity verification threshold, the identity authentication result is successful, otherwise it fails.

[0012] In an exemplary embodiment, during the communication between the client and the cloud server, the method also includes: retrieving corresponding transmission data and attaching a message authentication code to the transmission data; encrypting the communication between the client and the cloud server using the transport layer security protocol so that the sender and the receiver can decrypt the voiceprint data; retrieving the corresponding key management policy and applying the key management policy to the communication between the client and the cloud server, wherein the key management policy includes the generation, distribution, storage and periodic update of the key to ensure the security of the key.

[0013] According to one aspect of the present application, a large-scale voiceprint database management device based on cloud services includes: a voiceprint feature vector extraction module, which receives and stores the user's voiceprint data to a cloud distributed storage system, preprocesses the voiceprint data to extract the corresponding voiceprint feature vector, and stores the voiceprint feature vector in a distributed index structure; a verification feature vector determination module, which, when receiving an identity authentication request, obtains the corresponding voiceprint data to be verified, preprocesses the voiceprint data to be verified to obtain the corresponding voiceprint feature vector to be verified, and quickly retrieves the stored voiceprint feature vector that is most similar to the voiceprint feature vector to be verified in the distributed index structure to determine the corresponding verification feature vector; an identity authentication result determination module, which calculates the similarity score between the voiceprint feature vector to be verified and the verification feature vector, and determines the corresponding identity authentication result based on the similarity score.

[0014] According to one aspect of the present application, an electronic device includes at least one processor and at least one memory, wherein the memory stores computer-readable instructions; the computer-readable instructions are executed by one or more of the processors, so that the electronic device implements the large-scale voiceprint database management method based on cloud services as described above.

[0015] According to one aspect of the present application, a storage medium stores computer-readable instructions thereon, and the computer-readable instructions are executed by one or more processors to implement the large-scale voiceprint database management method based on cloud services as described above.

[0016] According to one aspect of the present application, a computer program product includes computer-readable instructions, which are stored in a storage medium. One or more processors of an electronic device read the computer-readable instructions from the storage medium, load and execute the computer-readable instructions, so that the electronic device implements the large-scale voiceprint database management method based on cloud services as described above.

[0017] The beneficial effects of the technical solution provided by this application are:

[0018] The user's voiceprint data is stored in a distributed storage system in the cloud. The elastic scaling capability of cloud computing can be used to process and store large-scale voiceprint data. The voiceprint data is preprocessed to extract key voiceprint feature vectors, which are easier to store and retrieve, reducing data redundancy. The extracted voiceprint feature vectors are stored in a distributed index structure, which supports efficient retrieval operations and can find the most similar stored voiceprint feature vectors in a short time. The identity verification result is determined based on the similarity score, thereby achieving high-precision identity recognition and reducing misjudgments and missed judgments. Through efficient storage, fast retrieval, high-precision matching, dynamic resource management, and strong security guarantees, the overall performance and security of large-scale voiceprint database management can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are only some embodiments of the present application, and those skilled in the art can obtain other drawings based on these drawings without creative work.

[0020] Figure 1 It is a schematic diagram of the implementation environment involved in this application;

[0021] Figure 2 is a flow chart showing a method for managing a large-scale voiceprint database based on cloud services according to an exemplary embodiment;

[0022] Figure 3 It is a structural block diagram of a large-scale voiceprint database management device based on cloud service according to an exemplary embodiment. DETAILED DESCRIPTION

[0023] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present application.

[0024] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present disclosure refers to the presence of the features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0025] As can be seen from the above, the related art still has the defect of low accuracy of large-scale voiceprint database management based on cloud services. To this end, the large-scale voiceprint database management method based on cloud services provided in this application can effectively improve the accuracy of large-scale voiceprint database management based on cloud services. Accordingly, the large-scale voiceprint database management method based on cloud services is applicable to a large-scale voiceprint database management device based on cloud services, and the large-scale voiceprint database management device based on cloud services can be deployed in electronic devices.

[0026] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0027] Figure 1 The schematic diagram is a schematic diagram of an implementation environment involved in a large-scale voiceprint database management method based on cloud services. The implementation environment includes a collection end and a service end.

[0028] See also Figure 2 The embodiment of the present application provides a large-scale voiceprint database management method based on cloud service, which is applicable to electronic devices, which may be Figure 1 The server side in the implementation environment is shown.

[0029] In the following method embodiments, for ease of description, the execution subject of each step of the method is taken as an electronic device as an example for illustration, but this does not constitute a specific limitation.

[0030] like Figure 2 As shown, the method may include the following steps:

[0031] S100, receiving and storing the user's voiceprint data into a cloud distributed storage system, preprocessing the voiceprint data, extracting the corresponding voiceprint feature vector, and storing the voiceprint feature vector in a distributed index structure.

[0032] Among them, users can record and upload voiceprint samples through the client (such as mobile applications, web pages, etc.). These voiceprint samples can be voice fragments, usually containing the user's specific pronunciation or phrases; the received voiceprint data is uploaded to the cloud distributed storage system.

[0033] By leveraging the elastic storage capabilities of the cloud, large-scale voiceprint data can be processed and stored, and distributed storage ensures data redundancy and reliability. Even if a node fails, it will not affect the operation of the overall system.

[0034] A series of preprocessing operations are performed on the voiceprint data, including noise reduction, frame segmentation, feature parameter extraction, etc., and feature vectors that can represent the characteristics of the voiceprint are extracted from the preprocessed voiceprint data. The feature vectors are more compact than the original voiceprint data, and are easier to quickly retrieve and match, which improves the processing speed of the system.

[0035] The extracted voiceprint feature vectors are stored in distributed index structures, such as hash tables, tree structures or other efficient indexing mechanisms. These index structures support fast search and retrieval. It should be pointed out here that the index structure itself is also distributed and can be stored and queried on multiple nodes, thereby improving the retrieval speed and the fault tolerance of the system. The distributed design supports high concurrent requests to ensure that the system can run stably during peak hours.

[0036] S110, when an identity authentication request is received, the corresponding voiceprint data to be verified is obtained, and the corresponding voiceprint feature vector to be verified is obtained by preprocessing the voiceprint data to be verified based on the latest voiceprint sample feature vector, and the corresponding verification feature vector is determined based on the rapid retrieval of the latest voiceprint sample feature vector distributed index structure that is most similar to the latest voiceprint sample feature vector to be verified.

[0037] Among them, when the user needs to authenticate his identity, he submits a request containing the voiceprint data to be verified through the client. The system obtains the voiceprint data to be verified from the request, usually a new voice segment, that is, the corresponding voiceprint data to be verified; the same preprocessing operations are performed on the voiceprint data to be verified, including noise reduction, framing, feature parameter extraction, etc., and the voiceprint feature vector to be verified is extracted from the preprocessed voiceprint data to be verified, which can ensure that the voiceprint feature vector to be verified is consistent with the stored voiceprint feature vector in processing method, thereby improving the accuracy of matching. At the same time, the preprocessing and feature vector extraction improve the efficiency of the subsequent matching process.

[0038] In addition, the distributed index structure is used to quickly retrieve the stored voiceprint feature vector that is most similar to the voiceprint feature vector to be verified. The similarity between the voiceprint feature vector to be verified and the stored voiceprint feature vector is calculated through a similarity algorithm (such as cosine similarity, Euclidean distance, etc.). The similarity calculation provides accurate matching results and improves the accuracy of identity authentication.

[0039] S120, calculating the similarity score between the latest voiceprint sample feature vector, the to-be-verified voiceprint feature vector and the verification feature vector, and determining the corresponding identity authentication result according to the similarity score between the latest voiceprint sample feature vector.

[0040] Among them, a similarity algorithm is used to calculate the similarity score between the voiceprint feature vector to be verified and the most similar stored voiceprint feature vector. According to the pre-set similarity threshold, it is judged whether the similarity score meets the authentication standard. If the score is higher than the threshold, the authentication is successful; otherwise, the authentication fails. Through the calculation of the similarity score and the threshold judgment, the system can accurately determine the identity authentication result.

[0041] Through efficient storage, preprocessing, feature vector extraction, distributed indexing, fast retrieval, and similarity calculation, efficient, accurate, and secure identity authentication is achieved. Each step is designed to optimize the system's performance and user experience while ensuring data security and privacy protection.

[0042] In the process of determining the corresponding identity authentication result according to the similarity score of the latest voiceprint sample feature vector, the method further includes:

[0043] S1201, retrieve a preset similarity verification threshold.

[0044] Among them, a similarity verification threshold is pre-set in the system. This threshold is determined based on the security requirements of the system and the actual test results. Usually, this threshold is set by the administrator during the system initialization or configuration stage.

[0045] S1202, comparing the similarity score of the latest voiceprint sample feature vector with the similarity verification threshold of the latest voiceprint sample feature vector.

[0046] The system has calculated the similarity score between the voiceprint feature vector to be verified and the stored voiceprint feature vector; the calculated similarity score is compared with the retrieved similarity verification threshold. The comparison operation is usually a simple numerical comparison, such as using operators such as greater than (>) and less than (<).

[0047] S1203, if the similarity score of the latest voiceprint sample feature vector is greater than the similarity verification threshold of the latest voiceprint sample feature vector, the identity authentication result is successful, otherwise it fails.

[0048] If the similarity score is greater than or equal to the preset similarity verification threshold, the system determines that the identity authentication is successful; if the similarity score is less than the preset similarity verification threshold, the system determines that the identity authentication fails. The system returns the identity authentication result to the client application, and may also include relevant status information (such as success / failure message).

[0049] The accuracy and security of the system are ensured by calling the preset similarity verification threshold, comparing the similarity score with the threshold, and determining the success or failure of identity authentication based on the comparison result. Each step is designed to provide a reliable identity authentication mechanism to ensure that only highly matched voiceprint feature vectors can pass the verification, thereby protecting the user's security and privacy.

[0050] When using the database system, the user's voiceprint data needs to be updated. The specific update process includes the following steps:

[0051] S1211, receiving a voiceprint data update request submitted by a user through a client application, and sending guidance information to the client based on the voiceprint data update request for the latest voiceprint sample feature vector to guide the user to record a new voiceprint sample.

[0052] Among them, the user triggers the voiceprint data update request through the client application (such as mobile application, web page, etc.), which can be a simple button or menu option. The user can choose to update his voiceprint data. The system receives the user's voiceprint data update request and prepares to start the update process.

[0053] The system generates and sends detailed guidance information to the client based on preset guidelines and best practices. This guidance information may include how to pronounce correctly, requirements for the recording environment, recording duration, etc. The client application displays this guidance information and guides the user to record the voiceprint sample according to the instructions.

[0054] Through detailed guidance information, it helps users record high-quality voiceprint samples, thereby improving the accuracy of subsequent processing and matching, ensuring that all users' voiceprint samples follow the same standards during the recording process, and reducing errors caused by different recording methods.

[0055] S1212, obtaining the latest voiceprint sample recorded by the user, preprocessing the latest voiceprint sample feature vector, and extracting the latest voiceprint sample feature vector.

[0056] Among them, the client uploads the latest voiceprint sample recorded by the user to the cloud server, and then performs a series of preprocessing operations on the newly recorded voiceprint sample, such as noise reduction, frame segmentation, feature parameter extraction, etc., and extracts the latest voiceprint sample feature vector from the preprocessed voiceprint sample; by extracting the feature vector, data redundancy is reduced, and storage efficiency and retrieval speed are improved.

[0057] S1213, storing the latest voiceprint sample feature vector in the cloud distributed storage system, updating the distributed index structure, and sending a notification of successful update to the user to require the user to authenticate to confirm the success of the update.

[0058] Among them, the latest voiceprint sample feature vector extracted is stored in the cloud distributed storage system, and the distributed index structure is updated accordingly, and the latest voiceprint sample feature vector is added to the index to ensure that the user's voiceprint data is always up to date, thereby improving the accuracy and reliability of identity authentication. The updated index structure supports fast search and ensures that the system can still run efficiently under high load.

[0059] In addition, the system sends a notification of successful update to the user, informing the user that the voiceprint data has been successfully updated. In order to confirm the integrity and security of the update process, the system requires the user to perform a brief identity verification. For example, the user may need to record a voice sample again, and the system will match it to confirm the success of the update.

[0060] Efficient, accurate and secure updating of voiceprint data is achieved through user-triggered voiceprint data update requests, detailed recording guidance, preprocessing and feature vector extraction, storage and index update, and final identity verification confirmation.

[0061] The above steps mentioned the update of the registered users. The voiceprint data is collected at each identity authentication or regularly. The method specifically performs the following steps:

[0062] S1221, obtain the user's voiceprint sample, retrieve the existing voiceprint model; compare the latest voiceprint sample feature vector voiceprint sample with the existing voiceprint model to determine whether there is a change.

[0063] Among them, at each identity verification or regularly (for example, monthly or quarterly), the system obtains the user's latest voiceprint sample through the client application, which can be done by the user's active upload or automatic triggering of the system, so as to ensure that the system's voiceprint model can reflect the user's latest voiceprint characteristics. A preliminary quality check will also be performed on the newly collected voiceprint samples to ensure that they meet the preset standards (such as clarity, duration, etc.). Through quality checks, it is ensured that the voiceprint samples used for updates have sufficient quality and consistency, thereby improving the accuracy of subsequent processing.

[0064] In addition, the existing voiceprint model of the user is retrieved from the cloud distributed storage system. This existing voiceprint model is generated based on the previously collected and processed voiceprint data. The newly collected voiceprint samples are preprocessed to extract the corresponding voiceprint feature vectors. The similarity score between the newly collected voiceprint feature vectors and the existing voiceprint model can be calculated by using a similarity algorithm (such as cosine similarity, Euclidean distance, etc.), and then based on a pre-set threshold, it is determined whether the similarity score is below a certain threshold. If it is below the threshold, it is considered that the voiceprint features have changed significantly.

[0065] S1222: If a change is detected, the existing voiceprint model is updated based on the incremental learning algorithm and the latest voiceprint sample feature vector voiceprint sample.

[0066] If a significant change in the voiceprint characteristics is detected, the system will start the incremental learning algorithm, which can update the existing model with new voiceprint samples without completely retraining the model. The newly collected voiceprint samples and their feature vectors are incorporated into the existing voiceprint model, and the model parameters are adjusted to adapt to the new voiceprint characteristics. The updated model needs to undergo a series of verification tests to ensure its accuracy and robustness.

[0067] By collecting users' voiceprint samples regularly or on demand, comparing them with existing voiceprint models, and updating the models using incremental learning algorithms when changes are detected, adaptive management and continuous optimization of the voiceprint database are achieved. Each step is designed to ensure the high accuracy, efficiency and adaptability of the system, thereby improving user experience and system reliability.

[0068] For the database in this application solution, corresponding data centers are deployed in multiple regions, including:

[0069] S1231, deploy data centers in different geographical areas and determine corresponding primary data centers and backup data centers.

[0070] Among them, multiple data centers are deployed in different geographical areas (such as different cities and countries). Each data center has the ability to process and store voiceprint data. And determine a primary data center and multiple backup data centers. The primary data center is responsible for the main data processing and service requests, while the backup data center serves as a backup in case of emergency. Through an efficient data synchronization mechanism (such as asynchronous replication or synchronous replication), ensure that the voiceprint data between all data centers is consistent.

[0071] S1222, monitor the status of the main data center of the latest voiceprint sample feature vector and the backup data center of all the latest voiceprint sample feature vectors. If an abnormality is detected, issue an alarm and initiate corresponding recovery measures.

[0072] The system continuously monitors the operating status of the primary data center and all backup data centers through monitoring tools and log analysis. The monitoring indicators include CPU usage, memory usage, disk I / O, network bandwidth, service response time, etc. Once an abnormal operating status of a data center is detected (such as excessive resource utilization, long response time, hardware failure, etc.), the system will immediately issue an alarm notification.

[0073] Alerts can be sent in a variety of ways, such as email, SMS, instant messaging, etc., so that administrators can respond in a timely manner. Recovery measures are based on the preset recovery plan. The system will automatically or manually start the corresponding recovery measures. These recovery measures may include restarting services, switching to backup nodes, resynchronizing data, etc.

[0074] S1233, when the main data center of the latest voiceprint sample feature vector fails, switch to the corresponding backup data center.

[0075] When a failure occurs in the primary data center (such as hardware failure, software crash, network interruption, etc.), the system will detect this through monitoring tools, and the system will automatically redirect service requests to the pre-configured backup data center. This process usually involves DNS switching, load balancer configuration updates, etc. During the switching process, the system will ensure that the data between the primary data center and the backup data center is synchronized to avoid data loss or inconsistency.

[0076] After the switch is complete, the system will perform a series of verification tests to ensure that the backup data center can provide services normally and that data integrity is not affected; the system can also send notifications to users to inform them that the service has been switched to the backup data center and confirm that the service has returned to normal.

[0077] By deploying data centers in different geographical areas, the high availability and disaster recovery capabilities of the system are ensured. Through continuous monitoring and automatic switching mechanisms, it is possible to quickly switch to the backup data center when a failure occurs in the primary data center, ensuring service continuity and data security.

[0078] For the voiceprint database management system, the method further performs the following steps:

[0079] S1241, obtaining the indicator data set corresponding to the database management system.

[0080] The system collects various performance indicators of the database management system regularly or in real time through monitoring tools (such as Prometheus, Grafana, CloudWatch, etc.). These indicators include but are not limited to: CPU usage: the current CPU utilization percentage; memory usage: the amount of memory currently used by the system; network bandwidth: the usage of network traffic, including upload and download rates; service request volume: the number of service requests received per unit time. By collecting these key indicators, the system can fully understand the operating status and load of the database management system.

[0081] S1242, detecting the latest voiceprint sample feature vector index data set, if it is detected that the load increases, adding computing resources to process additional requests; if the load decreases, determining the corresponding redundant resources.

[0082] Among them, the system analyzes the collected indicator data to determine the current load situation. For example, if the CPU utilization rate exceeds a certain threshold (such as 80%), or the service request volume increases significantly, the load is considered to have increased. When an increase in load is detected, the system automatically adds more computing resources (such as virtual machines, containers, server instances, etc.) to handle additional requests. This is usually done automatically through the cloud service provider's API. When a decrease in load is detected, the system identifies excess computing resources and releases these resources to save costs. For example, shutting down virtual machines or containers that are no longer needed.

[0083] Dynamically adjust resources according to actual load conditions to ensure that the system can maintain high performance under high load while saving resource costs under low load.

[0084] S1243, when a node failure of the database management system is detected, the corresponding failed node is determined, and the task corresponding to the failed node of the latest voiceprint sample feature vector is transferred to other healthy nodes.

[0085] The system continuously monitors the status of each node through monitoring tools. Once a node has an abnormality (such as response timeout, hardware failure, etc.), the system will detect this immediately. The system determines the specific faulty node and records its relevant information (such as IP address, node ID, etc.). The system automatically transfers the tasks on the faulty node to other healthy nodes. This is usually achieved through a load balancer or cluster management tools (such as Kubernetes, Docker Swarm, etc.). After the task transfer is completed, the system will perform a series of verification tests to ensure that the task runs normally on the new node and that data consistency is not affected. The system can send alarm notifications to administrators to inform them of the status of the faulty node and the results of the task transfer.

[0086] By executing steps S1241 to S1243, the high availability, elasticity and cost-effectiveness of the system are ensured by real-time monitoring of the performance indicators of the database management system, dynamically adjusting computing resources, and automatically transferring tasks when a node fails.

[0087] During the communication between the client and the cloud server, the method further includes:

[0088] S1251, retrieve the corresponding transmission data, and attach a message authentication code to the latest voiceprint sample feature vector transmission data.

[0089] The data to be transmitted is obtained from the client or cloud server, which may include voiceprint samples, identity verification requests, etc. A symmetric encryption algorithm (such as HMAC-SHA256) is used to generate a message authentication code (MAC) for the transmitted data. MAC is a fixed-length value calculated based on the transmitted data and a shared key.

[0090] S1252, using the transport layer security protocol to encrypt the communication between the client and the cloud server, so that the sender and the receiver can decrypt the voiceprint data.

[0091] Among them, a secure connection is established between the client and the cloud server through the TLS / SSL handshake process. This process includes exchanging certificates, negotiating encryption algorithms, and generating session keys. After the secure connection is established, all transmitted data is encrypted using the negotiated encryption algorithm. Common encryption algorithms include AES, RSA, etc. The encrypted data is transmitted over the network, and only the recipient with the correct key can decrypt the data.

[0092] S1253, retrieve the corresponding key management policy, and apply the latest voiceprint sample feature vector key management policy to the communication between the client and the cloud server.

[0093] Among them, the key management strategy includes key generation, distribution, storage and regular update to ensure the security of the key.

[0094] Use a strong random number generator to generate encryption keys. The length and complexity of the key should meet security standards, such as using a 256-bit AES key, and distribute the key through a secure channel (such as a public key infrastructure PKI). Ensure that the key is not intercepted during transmission, store the key securely in a protected environment, such as a hardware security module (HSM) or an encrypted database, and update the key regularly to prevent the key from being used for a long time and increasing the risk of being cracked. The update frequency can be set according to the security requirements of the system, such as once every three months.

[0095] By attaching a message authentication code to the transmitted data, encrypting it using the transport layer security protocol, and implementing a strict key management policy, the security and integrity of the communication between the client and the cloud server are ensured. Each step is designed to provide multi-level security protection to prevent data from being eavesdropped, tampered with, or forged during transmission, thereby protecting the user's data security and privacy.

[0096] The following is an embodiment of the device of the present application, which can be used to execute the large-scale voiceprint database management method based on cloud services involved in the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the method embodiment of the large-scale voiceprint database management method based on cloud services involved in the present application.

[0097] See also Figure 3 In the embodiment of the present application, a large-scale voiceprint database management device based on cloud service is provided, including but not limited to:

[0098] The voiceprint feature vector extraction module 200 receives and stores the user's voiceprint data in a cloud distributed storage system, pre-processes the latest voiceprint sample feature vector voiceprint data to extract the corresponding voiceprint feature vector, and stores the latest voiceprint sample feature vector voiceprint feature vector in a distributed index structure;

[0099] The verification feature vector determination module 210, when receiving an identity authentication request, obtains the corresponding voiceprint data to be verified, pre-processes the voiceprint data to be verified based on the latest voiceprint sample feature vector to obtain the corresponding voiceprint feature vector to be verified, and quickly retrieves the latest voiceprint sample feature vector stored most similar to the latest voiceprint sample feature vector to be verified based on the distributed index structure of the latest voiceprint sample feature vector to determine the corresponding verification feature vector;

[0100] The identity authentication result determination module 220 calculates the similarity score between the latest voiceprint sample feature vector, the to-be-verified voiceprint feature vector and the verification feature vector, and determines the corresponding identity authentication result according to the similarity score between the latest voiceprint sample feature vector.

[0101] In an exemplary embodiment, the device further includes but is not limited to:

[0102] A voiceprint data update request receiving module is used to receive a voiceprint data update request submitted by a user through a client application, and send guidance information to the client based on the voiceprint data update request of the latest voiceprint sample feature vector to guide the user to record a new voiceprint sample;

[0103] The latest voiceprint sample acquisition module is used to obtain the latest voiceprint sample recorded by the user, preprocess the latest voiceprint sample feature vector, and extract the latest voiceprint sample feature vector;

[0104] The latest voiceprint sample feature vector storage module is used to store the latest voiceprint sample feature vector in the cloud distributed storage system, update the distributed index structure, and send a notification of successful update to the user to require the user to authenticate to confirm the success of the update.

[0105] In an exemplary embodiment, the device further includes but is not limited to:

[0106] The voiceprint sample acquisition module obtains the user's voiceprint sample and calls the existing voiceprint model; compares the latest voiceprint sample feature vector voiceprint sample with the existing voiceprint model to determine whether there is a change;

[0107] The voiceprint model update module uses the incremental learning algorithm and the latest voiceprint sample feature vector to update the existing voiceprint model if a change is detected.

[0108] In an exemplary embodiment, the device further includes but is not limited to:

[0109] A data center deployment module is used to deploy data centers in different geographical areas and determine the corresponding primary data centers and backup data centers, wherein the voiceprint data between the latest voiceprint sample feature vector data centers are consistent;

[0110] The monitoring module is used to monitor the status of the main data center of the latest voiceprint sample feature vector and the backup data center of all the latest voiceprint sample feature vectors. If an abnormality is detected, an alarm is issued and corresponding recovery measures are initiated;

[0111] The data center switching module is used to switch to the corresponding backup data center when the main data center of the latest voiceprint sample feature vector fails.

[0112] In an exemplary embodiment, the device further includes but is not limited to:

[0113] The indicator data set acquisition module is used to obtain the indicator data set corresponding to the database management system, and the indicator data set includes CPU usage, memory usage, network bandwidth and service request volume;

[0114] The indicator data detection module is used to detect the latest voiceprint sample feature vector indicator data set. If an increase in load is detected, computing resources are added to process additional requests; if the load decreases, the corresponding redundant resources are determined;

[0115] The task transfer module, when a node failure of the database management system is detected, determines the corresponding failed node and is used to transfer the task corresponding to the failed node of the latest voiceprint sample feature vector to other healthy nodes.

[0116] In an exemplary embodiment, the device further includes but is not limited to:

[0117] Retrieve the preset similarity verification threshold;

[0118] Compare the similarity score of the latest voiceprint sample feature vector with the similarity verification threshold of the latest voiceprint sample feature vector;

[0119] If the similarity score of the latest voiceprint sample feature vector is greater than the similarity verification threshold of the latest voiceprint sample feature vector, the identity authentication result is successful, otherwise it fails.

[0120] In an exemplary embodiment, the device further includes but is not limited to:

[0121] The transmission data retrieval module is used to retrieve the corresponding transmission data and append a message authentication code to the latest voiceprint sample feature vector transmission data;

[0122] An encryption module, used to encrypt the communication between the client and the cloud server using the transport layer security protocol, so that the sender and the receiver can decrypt the voiceprint data;

[0123] The key management policy retrieval module is used to retrieve the corresponding key management policy and apply the latest voiceprint sample feature vector key management policy to the communication between the client and the cloud server. The key management policy includes the generation, distribution, storage and regular update of the key to ensure the security of the key.

[0124] It should be noted that the large-scale voiceprint database management device based on cloud services provided in the above embodiments only uses the division of the above-mentioned functional modules as an example when performing large-scale voiceprint database management based on cloud services. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the large-scale voiceprint database management device based on cloud services will be divided into different functional modules to complete all or part of the functions described above.

[0125] In addition, the large-scale voiceprint database management device based on cloud services and the large-scale voiceprint database management method based on cloud services provided in the above embodiments belong to the same concept, and the specific way in which each module performs operations has been described in detail in the method embodiments and will not be repeated here.

[0126] According to one aspect of the present application, an electronic device includes at least one processor and at least one memory, wherein the memory stores computer-readable instructions; the computer-readable instructions are executed by one or more of the processors, so that the electronic device implements the large-scale voiceprint database management method based on cloud services as described above.

[0127] In addition, a storage medium is provided in an embodiment of the present application, on which computer-readable instructions are stored, and the computer-readable instructions are executed by one or more processors to implement the large-scale voiceprint database management method based on cloud services as described above.

[0128] In an embodiment of the present application, a computer program product is provided, which includes computer-readable instructions. The computer-readable instructions are stored in a storage medium. One or more processors of an electronic device read the computer-readable instructions from the storage medium, load and execute the computer-readable instructions, so that the electronic device implements the large-scale voiceprint database management method based on cloud services as described above.

[0129] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0130] The above description is only a partial implementation method of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A large-scale voiceprint database management method based on cloud services, characterized in that: include: Receive and store the user's voiceprint data in a cloud distributed storage system, pre-process the voiceprint data, extract the corresponding voiceprint feature vector, and store the voiceprint feature vector in a distributed index structure; When an identity authentication request is received, the corresponding voiceprint data to be verified is obtained, the corresponding voiceprint feature vector to be verified is obtained by preprocessing the voiceprint data to be verified, and the corresponding verification feature vector is determined based on the quick retrieval of the stored voiceprint feature vector that is most similar to the voiceprint feature vector to be verified in the distributed index structure; The similarity score between the voiceprint feature vector to be verified and the verification feature vector is calculated, and the corresponding identity verification result is determined according to the similarity score.

2. The method according to claim 1, characterized in that The method also includes: Receiving a voiceprint data update request submitted by a user through a client application, and sending guidance information to the client based on the voiceprint data update request to guide the user to record a new voiceprint sample; Obtain the latest voiceprint sample recorded by the user, preprocess the latest voiceprint sample, and extract the latest voiceprint sample feature vector; The latest voiceprint sample feature vector is stored in the cloud distributed storage system, and the distributed index structure is updated. A notification of successful update is sent to the user to require the user to authenticate to confirm the success of the update.

3. The method according to claim 2, characterized in that Collecting voiceprint data at each identity verification or periodically, the method also includes: Obtain a voiceprint sample of the user and retrieve an existing voiceprint model; compare the voiceprint sample with the existing voiceprint model to determine whether there is a change; If a change is detected, the existing voiceprint model is updated based on the incremental learning algorithm and the voiceprint sample.

4. The method according to claim 1, characterized in that The method also includes: Deploy data centers in different geographical areas and determine corresponding primary data centers and backup data centers, wherein the voiceprint data between the data centers remain consistent; Monitor the status of the primary data center and all the backup data centers, and if an abnormality is detected, issue an alarm and initiate corresponding recovery measures; When the primary data center fails, it switches to the corresponding backup data center.

5. The method according to claim 1, characterized in that The method also includes: Obtain the indicator data set corresponding to the database management system, including CPU usage, memory usage, network bandwidth, and service request volume; The indicator data set is detected, and if an increase in load is detected, computing resources are added to process additional requests; if the load is reduced, corresponding redundant resources are determined; When a node failure of the database management system is detected, the corresponding failed node is determined, and the tasks corresponding to the failed node are transferred to other healthy nodes.

6. The method according to claim 1, characterized in that In the process of determining the corresponding identity authentication result according to the similarity score, the method further includes: Retrieve the preset similarity verification threshold; comparing the similarity score with the similarity verification threshold; If the similarity score is greater than the similarity verification threshold, the identity authentication result is successful, otherwise it fails.

7. The method according to claim 1, characterized in that During the communication between the client and the cloud server, the method further includes: Retrieving corresponding transmission data, and adding a message authentication code to the transmission data; Use the Transport Layer Security protocol to encrypt the communication between the client and the cloud server, so that the sender and receiver can decrypt the voiceprint data; Retrieve the corresponding key management policy and apply the key management policy to the communication between the client and the cloud server, wherein the key management policy includes the generation, distribution, storage and regular update of the key to ensure the security of the key.

8. A large-scale voiceprint database management device based on cloud service, characterized in that: include: A voiceprint feature vector extraction module receives and stores the user's voiceprint data in a cloud distributed storage system, pre-processes the voiceprint data to extract the corresponding voiceprint feature vector, and stores the voiceprint feature vector in a distributed index structure; A verification feature vector determination module, when receiving an identity authentication request, obtains corresponding voiceprint data to be verified, performs preprocessing based on the voiceprint data to be verified to obtain a corresponding voiceprint feature vector to be verified, and quickly retrieves the stored voiceprint feature vector that is most similar to the voiceprint feature vector to be verified based on the distributed index structure to determine the corresponding verification feature vector; The identity authentication result determination module calculates the similarity score between the voiceprint feature vector to be verified and the verification feature vector, and uses the similarity score to determine the corresponding identity authentication result.

9. An electronic device, characterized in that: include: at least one processor and at least one memory, wherein: The memory has computer-readable instructions stored thereon; The computer-readable instructions are executed by one or more of the processors, so that the electronic device implements the cloud service-based large-scale voiceprint database management method as described in any one of claims 1 to 7.

10. A storage medium having computer-readable instructions stored thereon, characterized in that: The computer-readable instructions are executed by one or more processors to implement the cloud service-based large-scale voiceprint database management method according to any one of claims 1 to 7.

Images