Digital duplicate model generation method and device

Abstract

The invention discloses a digital duplicate model generation method and device, and the method comprises the steps: obtaining personal information which is submitted by a user and is used for generating a digital duplicate, and carrying out the compliance verification of the personal information; if the personal information passes the verification, generating at least one newly added clone function option in a user interface; in response to selection of the user on the cloning function option, associating the personal information passing verification with a cloning type corresponding to the option selected by the user, and packaging the personal information into a structured cloning request; and obtaining the cloning request, distributing the cloning request to the corresponding digital cloned agent, and generating a digital cloned model matched with the personal information of the user based on the key parameters. The interaction efficiency is improved through dynamic UI updating, the cloning accuracy is guaranteed through multi-dimensional information verification, the response time is shortened through an automatic proxy link, meanwhile, a compliant data security mechanism is established, and efficient, safe and personalized cloning service is achieved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method and apparatus for generating digital clone models. Background Technology

[0002] With the rapid development of artificial intelligence and digital human technology, the demand for services that generate personalized digital avatars based on users' personal images and voice characteristics is growing. However, existing digital avatar generation systems still have significant technical bottlenecks in their core processing flow, which severely restricts their response efficiency and scalability.

[0003] In current technical solutions, the backend processing flow of a system often cannot be fully automated after a user submits a clone generation request and related personal information. Specifically, the distribution and scheduling of user requests, as well as the identification and interpretation of cloning parameters (such as image features, voice parameters, and text content) contained in the request, all require manual intervention. This manual intervention includes not only assigning tasks to the corresponding processing units, but also having technical personnel understand and transform unstructured user requests to configure the input parameters required by the clone generation model. Due to the inherent delay of manual processing, the system's response time from receiving a request to starting execution is significantly extended, often requiring users to wait for several minutes or even longer, resulting in a poor user experience. Secondly, there is an upper limit to manual processing capabilities, severely limiting the overall concurrent processing capacity of the system and making it difficult to support high-concurrency user requests. Furthermore, this process is highly dependent on the operation of specific personnel, posing a risk of inconsistent processing standards and the introduction of human error, affecting the consistency of digital clone generation quality. Summary of the Invention

[0004] To address the aforementioned issues, this application provides a method and apparatus for generating digital clone models.

[0005] The embodiments of this application disclose the following technical solutions: The first aspect of this application provides a method for generating a digital clone model, including: Obtain personal information submitted by the user for generating a digital avatar, the personal information including at least one type of data used to characterize the user's voice and / or user appearance; The personal information will be verified for compliance. If the personal information passes the verification, at least one new clone function option will be generated in the user interface; In response to the user's selection of the cloning function option, the verified personal information is associated with the cloning type corresponding to the user's selected option and encapsulated into a structured cloning request. The cloning request is obtained and distributed to the corresponding digital clone agent, so that the digital clone agent can parse the key parameters required to generate the digital clone from the cloning request; Based on the key parameters, a digital avatar model matching the user's personal information is generated. The digital avatar model is a personalized digital human model generated based on the user's voice and / or image.

[0006] In one possible implementation, before obtaining the personal information submitted by the user for generating the digital avatar, the method further includes: In response to the user's request to generate a digital clone model, the initial function options are displayed in the user interface; Based on the initial function options selected by the user, guide information is generated and displayed on the user interface so that the user can enter personal information on the user interface based on the guide information.

[0007] In one possible implementation, the compliance verification of personal information includes at least one of the following: The system uses OCR to identify the name in the ID photo and matches it with the name submitted by the user, and uses a face liveness detection model to determine whether the ID photo is a real face. The sound features in the sound samples are extracted using the MFCC algorithm, and the integrity of the sound features is determined. The CTC text matching model is used to compare the text identified from the sound samples with the corresponding text submitted by the user.

[0008] In one possible implementation, the step of obtaining the cloning request and distributing it to the corresponding digital clone agent, whereby the digital clone agent parses the key parameters required to generate the digital clone from the cloning request, including: The AI ​​agent listens to the message queue and captures the encapsulated clone requests; Based on the clone type encapsulated in the clone request and the load status of each digital clone agent, the clone request is distributed to the corresponding digital clone agent so that the digital clone agent runs a structured data parsing algorithm to extract the image parameters, sound parameters and text parameters from the clone request as the key parameters required to generate the digital clone.

[0009] In one possible implementation, generating a digital avatar model that matches the user's personal information based on the key parameters includes: Based on the sound parameters parsed from the cloning request, a digital clone audio that matches the user's voice timbre is generated; based on the image parameters parsed from the cloning request, a digital clone image is generated, and a lip-sync algorithm is used to match the lip movements of the digital clone image with the time frame of the digital clone audio to obtain a digital clone model.

[0010] In one possible implementation, associating the verified personal information with the clone type corresponding to the user's selected option and encapsulating it into a structured clone request includes: Sensitive data in the personal information is encrypted, and a corresponding unique request identifier is generated; The encrypted personal information, the unique identifier of the request, the clone type corresponding to the user's selected option, and the user authorization information are jointly encapsulated into a structured clone request. The user authorization information is used to characterize the scope and validity period of the user's authorization for the use of personal information.

[0011] In one possible implementation, the method further includes: The encrypted structured request is persistently stored in a distributed database, and user authorization information, including the scope and validity period, is stored through a consortium blockchain, whose nodes include system servers and user terminals.

[0012] In one possible implementation, generating at least one new clone function option in the user interface includes: The backend service sends a dynamic update command to the user interface. The dynamic update command contains configuration information for the new option, so that the user interface can add the new clone function option to the user interface using a component-based rendering method based on the dynamic update command.

[0013] In one possible implementation, the method further includes: The audio files and preview video of the digital clone model are packaged into a download package, and a download link for the download package is generated; The download link is sent to the user interface, which then displays audio download options, a preview area, and operation instructions based on the download link. The lip movements of the digital avatar shown in the preview video are synchronized with the audio content.

[0014] A second aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the digital clone model generation method as described in the first aspect above.

[0015] A third aspect of this application provides a computer program product that, when run on a computer, executes the digital clone model generation method as described in the first aspect above.

[0016] A fourth aspect of this application provides a computer-readable storage medium storing instructions that, when executed on a terminal device, cause the terminal device to perform the digital clone model generation method as described in the first aspect above.

[0017] Compared with the prior art, this application has the following beneficial effects: After the user submits information and passes compliance verification, the system first intelligently guides the user to make selections by dynamically updating the user interface. Then, the verified information and the user's selections are encapsulated into a structured clone request, achieving machine-readable conversion between user intent and generated parameters. This transforms unstructured requirements that previously required manual interpretation into standardized processing instructions. Furthermore, by automatically acquiring and distributing the clone request to the corresponding digital avatar agent, an automated task distribution chain is established between the AI ​​agent and the digital avatar agent, completely replacing the manual request scheduling stage. The agent automatically parses key parameters, ultimately generating the digital avatar model. This process, with its automated closed loop of structured encapsulation, automatic distribution, and agent parsing, replaces the original manual distribution and parameter interpretation stages, significantly shortening system response time and greatly improving processing efficiency and concurrency capabilities. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart illustrating a method for generating a digital clone model provided in this application embodiment; Figure 2 This is a schematic diagram of a digital clone model generation system provided in an embodiment of this application. Detailed Implementation

[0020] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0021] To facilitate understanding of the technical solutions provided in the embodiments of this application, the terms involved in the embodiments of this application will be explained below.

[0022] A user interface (UI) is the medium or platform that connects users with systems (such as software, applications, websites, and smart devices) for interaction. Its core function is to enable two-way information exchange between users and machines: users issue commands to the system through the UI, and the system then uses the UI to provide feedback on processing results, data, or status to the user.

[0023] To facilitate understanding of the technical solutions provided in the embodiments of this application, the background technology involved in the embodiments of this application will be described below.

[0024] As mentioned earlier, existing technologies suffer from the following drawbacks: Existing systems often have preset function options that cannot be dynamically expanded based on user-submitted information. For example, after a user uploads an ID photo, they must manually search for the identity-related cloning function, resulting in low operational efficiency. Existing solutions only collect single information about the user's image or voice, failing to verify the consistency between the voice sample and the corresponding text, or the authenticity of the ID photo. This leads to a high error rate of mismatch between the digital clone's voice and text, or between the image and the identity. There is insufficient process automation: after a user submits a request, manual intervention is required to handle request distribution and cloning parameter interpretation, making it impossible to automate the AI-assisted digital clone process, resulting in long response delays. Furthermore, compliance and security are lacking: no standardized encrypted storage and authorization mechanism has been established for sensitive information such as user ID photos and voice samples, posing a risk of personal information leakage and failing to mitigate the ethical risks of digital clones being used for identity forgery.

[0025] To address the aforementioned issues, this application's embodiments firstly, after the user passes compliance verification, the system automatically and dynamically generates new function options in the user interface, allowing the user to quickly select the target function without manual searching, greatly simplifying the operation path. Secondly, by performing multiple automated verifications on the user-submitted ID photos, voice samples, and corresponding text, the authenticity and consistency of the information source are ensured, fundamentally improving the accuracy of the cloning model. Furthermore, by encapsulating user requests and clone types into structured data and establishing an automated request distribution and parsing mechanism, fully automated intelligent processing from task issuance to parameter extraction is achieved, effectively shortening response time. Finally, by combining data encryption with blockchain notarization, sensitive information is protected while a trusted traceability system is built through tamper-proof authorization records, comprehensively ensuring data security and compliance. This solution, through the synergistic effect of the above technical means, achieves a simultaneous improvement in the efficiency, quality, and security of digital clone generation.

[0026] It should be noted that the digital avatar model generation method, system, device, product, and medium provided in this application can be applied to the field of computer technology, as well as the field of artificial intelligence and user interface (UI) interaction technology. The above are merely examples and do not limit the application areas of the digital avatar model generation method, system, device, product, and medium provided in this application. Furthermore, the embodiments of this application may not limit the executing entity of digital avatar model generation. For example, the digital avatar model generation method of this application embodiment can be applied to data processing devices such as terminal devices or servers. The terminal device can be an electronic device such as a computer or a personal digital assistant (PDA). The server can be a standalone server, a cloud server, or a cluster server composed of multiple servers.

[0027] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0028] See Figure 1 , Figure 1 A flowchart of a digital clone model generation method provided in this application embodiment specifically includes: S101. Obtain the personal information submitted by the user for generating the digital clone.

[0029] Step S101 is the starting point and data collection stage of the entire digital avatar model generation process. Its core objective is to guide users to submit their personal information, providing a data foundation for subsequent verification and generation. The personal information used to generate the digital avatar is key data characterizing the user's unique biometrics, including at least one of the following: data characterizing the user's voice, such as a voice sample (sound sample) of specific content recorded by the user; and data characterizing the user's appearance, such as a passport photo with a clear face uploaded by the user.

[0030] In one possible implementation, before obtaining the personal information submitted by the user for generating the digital clone, the method further includes: in response to the user's request to generate the digital clone model, displaying initial function options on the user interface; and generating guidance information based on the initial function options selected by the user and displaying it on the user interface so that the user can input personal information on the user interface based on the guidance information.

[0031] First, the system needs to guide users to clarify their needs and submit the correct data. To this end, before directly obtaining personal information, the system will provide pre-interaction guidance: the user initiates a digital avatar generation request to the system through a terminal (such as a webpage or app), for example, asking a question like "How do I generate my own personalized digital avatar?" Upon receiving the request, the system's user interaction module will respond immediately, dynamically displaying five initial function options in a carousel format on the user interface, such as "Entertainment-related digital avatar generation," "Education-related digital avatar generation," "Customer service-related digital avatar generation," "Personal voice cloning," and "Customized image cloning."

[0032] This display method uses the CSS3 animation property combined with transform: translateX() to achieve smooth scrolling animation, rotating at a rate of 1 frame per second, thus clearly and vividly presenting all options within the limited screen space. At the same time, through media queries (@media) technology, the spacing and layout of the options will automatically adjust according to the screen resolution of different terminals (such as mobile phones, tablets, and computers), ensuring the best visual and interactive experience on any device.

[0033] Once a user browses and selects an initial feature option (such as "Personal Voice Cloning"), the system generates and displays targeted guidance information based on that selection. This guidance clearly informs the user what specific data needs to be prepared and submitted, as well as the data format requirements. With the assistance of this guidance, the user begins to enter personal information into the user interface to generate their digital avatar.

[0034] For example, users can submit four types of core information through the user interaction module based on the prompts in the guidance information: Basic Information: The user's real name, submitted as a string with a length limit of 1 to 20 characters.

[0035] ID photo: The photo must be in JPG or PNG format, with a resolution of at least 300dpi, and must contain a complete and clear face for subsequent image cloning.

[0036] Audio samples: must be WAV format audio files, with a duration of 3 to 10 seconds, no background noise, a standard sampling rate of 44.1kHz, and mono.

[0037] Voice-to-Text: Submit text that is completely identical to the content of the voice sample. For example, if the content of the voice sample is "Hello, I am user XXX", then the submitted text must strictly match it. This text will be used for subsequent voice-text consistency checks and speech synthesis for digital avatars.

[0038] S102. Perform compliance verification on the personal information.

[0039] Step S102 is a crucial quality control step in the entire digital avatar model generation process. Its core objective is to ensure that the personal information submitted by the user in step S101 is authentic, complete, and consistent, thereby guaranteeing the accuracy, security, and reliability of the final digital avatar from the source. This step can be executed by an information verification module that integrates multiple advanced algorithms to perform automated, multi-dimensional compliance verification on four types of core information.

[0040] Specifically, verification can include at least one of the following methods: Firstly, the authenticity verification of ID photos involves rigorously reviewing the ID photos submitted by users. This process combines OCR (Optical Character Recognition) technology with a face liveness detection model. The OCR algorithm automatically identifies and extracts the name information from the ID photo image, then precisely compares it with the user's previously submitted "basic information" (the user's real name). If the match rate reaches or exceeds 95%, the name is considered to have passed the verification. Simultaneously, a face liveness detection model (such as a model trained based on the MTCNN architecture combined with the Triplet Loss loss function) performs deep analysis on the image to determine if it is a genuine photograph of a live face, rather than a photocopy or an AI-generated image. If the model's liveness confidence score is not lower than 80 points, the ID photo authenticity verification is considered passed.

[0041] Secondly, regarding the integrity verification of the audio samples, the system employs the MFCC (Mel-frequency cepstral coefficients) algorithm for in-depth analysis of user-submitted audio samples. This algorithm can extract key acoustic features from the audio samples, such as fundamental frequency, speech rate, and spectrum. By analyzing the continuity of these feature parameters, the system can determine whether the audio sample has serious missing or interrupted components. Typically, if the missing feature parameter ratio is less than 5%, the audio sample is considered complete and valid, and has passed the integrity verification.

[0042] Thirdly, the consistency verification between voice and text is a crucial step in ensuring that the digital avatar can correctly read the specified text. The user-submitted voice sample is input into a pre-trained CTC (Connection-Temporal Classification) text matching model. This model identifies the audio content and outputs the identified text. Subsequently, the system compares this identified text character by character with the user-submitted "voice-text correspondence." If the character matching rate is not less than 98%, the voice and text content are considered highly consistent, and the verification passes.

[0043] The three checks mentioned above together form a rigorous verification chain. The system will integrate all verification results: only when all verification items pass successfully will the process proceed to the next step, S103. If any verification fails, the system will immediately interrupt the process and return specific and clear error messages to the user through the user interaction module (e.g., "ID photo recognition failed, please upload a real and clear ID photo" or "Audio sample does not match the text, please re-record"), guiding the user to resubmit or correct the relevant information, thus forming an efficient feedback loop.

[0044] S103. If the personal information passes the verification, at least one new clone function option is generated in the user interface.

[0045] Once the information verification module in the previous step (S102) confirms that all personal information submitted by the user has passed compliance verification, it will immediately send a "verification passed" signal to the UI dynamic update module. This signal will trigger a seamless interface upgrade process without manual intervention. S103 successfully transforms the complex verification results completed in the backend into intuitive and immediate changes to the front-end user interface. By unlocking new features upon successful verification, it not only greatly improves interaction efficiency, eliminating the need for users to manually search or wait, but also constructs a clearly guided and smooth automated service process.

[0046] In one possible implementation, generating at least one new clone function option in the user interface includes: sending a dynamic update instruction to the user interface via a backend service, the instruction containing configuration information of the new option, so that the user interface adds the new clone function option to the user interface using a component-based rendering method based on the dynamic update instruction.

[0047] Once the information verification module confirms that all verifications have passed, it immediately sends a "verification passed" signal to the UI dynamic update module. Upon receiving this signal, the backend service generates a specific "add option instruction" and actively and in real-time pushes it to the frontend user interface via the WebSocket full-duplex communication protocol. This instruction is a structured data packet that explicitly contains the core configuration information for the new option: for example, the option name (e.g., "clone by ID card"), the icon resource path (e.g., / static / icons / id-clone.png), and the crucial trigger permission (marked as "visible only to verified users"). This ensures that the new feature is only available to eligible users.

[0048] When the front-end user interface receives instructions pushed from the back-end via WebSocket, it does not refresh the entire page like a traditional webpage. Instead, it leverages the reactivity and componentization capabilities of modern front-end frameworks (such as Vue.js) to dynamically update the interface. Specifically, the front-end calls Vue's `this.$set` method or similar reactive APIs to dynamically add the new options described in the instructions as complete UI components to the existing user interface data structure.

[0049] Newly added option components are rendered to the user immediately and seamlessly integrated into the layout of existing options (e.g., the initial five options). The entire process is smooth, with no page refreshes or flickering. More importantly, the visual style of the newly generated options, including but not limited to rounded corners, shadows, and hover effects, remains completely consistent with the original options, ensuring interface unity and aesthetics. This makes users feel that the new features are an integral part of the interface, rather than something awkwardly added on.

[0050] S104. In response to the user's selection of the cloning function option, the verified personal information is associated with the cloning type corresponding to the user's selected option, and encapsulated into a structured cloning request.

[0051] Responding to the user's selection behavior, the system clarifies the user's cloning intent. It precisely maps the user's selected UI option (such as "clone by ID document") to a specific, machine-readable cloning type. Subsequently, the system logically binds this cloning type to the personal information (including encrypted ID photo, processed voice feature parameters, and corresponding text) that has passed rigorous verification in step S102. This association is crucial, ensuring that subsequent processing flows clearly know "whose data" and "what type of cloning task" to perform. After completing the association between intent and data, the system's request processing module encapsulates these elements into a structured cloning request. This process typically organizes all information into a standard data exchange format, such as JSON.

[0052] In one possible implementation, associating the verified personal information with the clone type corresponding to the user's selected option and encapsulating it into a structured clone request includes: encrypting sensitive data in the personal information and generating a corresponding unique request identifier; encapsulating the encrypted personal information, the unique request identifier, the clone type corresponding to the user's selected option, and user authorization information together into a structured clone request, wherein the user authorization information is used to characterize the scope and validity period of the user's authorization for the use of personal information.

[0053] To ensure the absolute safety of sensitive personal information, the system implements high-strength security protection for the packaged requests. It employs the RSA asymmetric encryption algorithm to specifically encrypt core sensitive data within the request (such as ID photo images and voice feature parameters). The asymmetric encryption characteristic means that the encryption process uses a public key, while decryption requires a unique private key, effectively preventing malicious interception and cracking of data during transmission and storage. After encryption, the system generates a unique 32-bit hash value as the request ID, which acts like a unique tracking number for the "cloned order," used for subsequent tracking and identification throughout the entire process.

[0054] After encryption is complete, two parallel persistence operations are performed: First, distributed storage stores the encrypted, fully structured request in a distributed database (such as MongoDB). This distributed architecture ensures high data availability and resilience. Second, blockchain notarization: simultaneously, the system notifies users of the operation's authorization information (e.g., "Authorization is only for generating a personal digital avatar, valid for 7 days") via a consortium blockchain. This consortium blockchain's nodes include system servers and user terminals, jointly maintaining the ledger. Once the authorization record is written to the blockchain, it becomes an immutable and traceable electronic certificate, ensuring the authenticity of user authorization from both technical and legal perspectives and effectively preventing potential abuse disputes.

[0055] In summary, S104 not only completes the conversion from user intent to machine instructions, but also constructs a secure, trustworthy, and auditable data processing environment through encryption and blockchain technology. It transforms each user's cloning request into a strictly protected digital contract, providing secure and standardized input for subsequent automated generation.

[0056] S105. Obtain the cloning request and distribute the cloning request to the corresponding digital clone agent, so that the digital clone agent can parse out the key parameters required to generate the digital clone from the cloning request.

[0057] In one possible implementation, obtaining the clone request and distributing it to the corresponding digital avatar agent, whereby the digital avatar agent parses the key parameters required for generating the digital avatar from the clone request, includes: listening to the message queue through an AI agent to capture the encapsulated clone request; and distributing the clone request to the corresponding digital avatar agent based on the clone type encapsulated in the clone request and the load status of each digital avatar agent, so that the digital avatar agent runs a structured data parsing algorithm to extract image parameters, sound parameters, and text parameters from the clone request as the key parameters required for generating the digital avatar.

[0058] A dedicated AI agent proactively acquires tasks. A high-frequency, timed task (e.g., executing every 100 milliseconds) continuously monitors the "pending requests" channel in the message queue. Upon detecting an encrypted clone request saved in step S104, the AI ​​agent immediately retrieves it. Subsequently, the AI ​​agent decrypts the request using a private key paired with the encryption public key, recovering readable structured data. After decryption, the AI ​​agent precisely extracts two key pieces of information from the request: the clone type (e.g., "clone by ID document") and the core parameters required for generation.

[0059] Once a clear clone type is identified, the AI ​​agent acts as a scheduler, distributing decrypted requests to the corresponding digital clone agents (i.e., processing modules specifically responsible for executing clone generation tasks) based on that type. To maximize system resource utilization and avoid overloading individual nodes, a load balancing algorithm (such as round-robin) is employed in the distribution process. This algorithm automatically detects the current load status of each digital clone processing node and prioritizes assigning new tasks to idle nodes, thereby ensuring the system has high concurrency processing capabilities and can quickly respond to requests from multiple users.

[0060] After the request is successfully distributed to the target digital avatar agent, the agent runs a structured data parsing algorithm to perform a final interpretation of the request. It can accurately extract several key parameters necessary for generating the digital avatar from the request, primarily including: Image parameters: For example, 68 facial feature points extracted from ID photos using a face recognition model (such as MTCNN).

[0061] Sound parameters: such as sound feature vectors extracted using the MFCC algorithm.

[0062] Text parameters: The corresponding text submitted by the user that needs to be read aloud by the digital clone.

[0063] It can also include cloning rules: that is, cloning based on ID documents, for example, ensuring that the image is ≥95% consistent with the ID photo.

[0064] S106. Based on the key parameters, generate a digital clone that matches the user's personal information.

[0065] This involves using the analyzed key parameters to synthesize a digital avatar that highly matches the user in terms of timbre, image, and expression.

[0066] In one possible implementation, generating a digital clone that matches the user's personal information based on the key parameters includes: Based on the sound parameters parsed from the cloning request, a digital clone audio that matches the user's voice timbre is generated; based on the image parameters parsed from the cloning request, a digital clone image is generated, and a lip-sync algorithm is used to match the lip movements of the digital clone image with the timing frames of the digital clone audio.

[0067] The digital avatar agent inputs the parsed audio parameters into an advanced VITS TTS (text-to-speech) model. To achieve the optimal balance between voice similarity and speech naturalness, the "clone similarity" parameter in the model is preset to 92%. Based on this setting and the text parameters provided by the user, the model generates a digital avatar audio that is highly consistent with the user's original voice timbre. The output audio is in MP3 format and encoded using a high bitrate of 320kbps to ensure excellent sound quality, making the digital avatar's speech sound both realistic and natural.

[0068] To enable the digital avatar to speak realistically, the system performs sophisticated image-audio timing matching. First, based on facial features extracted from an ID photo, a corresponding 3D digital avatar is generated. Then, a dedicated 3D model lip-sync algorithm precisely matches the avatar's lip movements with the timing frames of the generated audio. This algorithm performs dynamic calibration at a high frequency of once every 0.5 seconds to ensure accurate lip-sync for both vowel and consonant pronunciations, achieving a highly realistic lip-reading synchronization effect.

[0069] In one possible implementation, after the generation and synchronization are complete, the digital avatar model generation and feedback module securely delivers the final result to the user. It packages the encrypted digital avatar audio and the image-audio matching preview video (which can be a 10-second MP4 file) into a complete download package. The system simultaneously generates a time-sensitive encrypted download link, for example, valid for 24 hours, and displays the final download interface to the user through the user interaction module. This interface clearly includes an audio download button, a preview video playback window, and necessary information, guiding the user to securely obtain and view the final effect of their personalized digital avatar.

[0070] In one possible implementation, this methodology also includes a delivery phase, the core objective of which is to ensure that users can securely and conveniently access the generated results and experience a highly consistent "image-sound-text" fusion effect. This step focuses on the final closed loop of the user experience, presenting the complete technical results of the preceding steps to the user.

[0071] When a user clicks the "Download" button on the download page, this action triggers the system to initiate a secure file transfer mechanism. The system uses an AES-256 encryption algorithm to establish a secure link for transmitting the download package containing digitally cloned audio and preview video. AES-256, as a high-strength symmetric encryption algorithm, ensures the confidentiality and integrity of data during transmission, effectively preventing the theft or tampering of sensitive user data during the download process.

[0072] After the download package is successfully transferred to the user's local device, the user performs an unzip operation. After unzipping, the user can open and view the final generated result. At this point, the user will see a digital 3D avatar generated based on their own ID photo. This avatar will simultaneously use a cloned voice sample highly consistent with their original voice, smoothly reading aloud the previously submitted "voice-corresponding text." This display ultimately achieves the high consistency of the "image-voice-text" triple features pursued by the system. Users can intuitively verify: Image authenticity: The digital avatar is based on the user's real ID photo. Voice similarity: The digital avatar's voice reproduces the user's own voice characteristics. Accuracy of expression: The content read by the digital avatar perfectly matches the text specified by the user.

[0073] The above are some specific implementations of the digital clone model generation method provided in the embodiments of this application. Based on this, this application also provides a corresponding digital clone model generation system. The system provided in the embodiments of this application will be described below from the perspective of functional modularity. See also... Figure 2 , Figure 2 This is a schematic diagram of a digital clone model generation system provided in an embodiment of this application.

[0074] The user interaction module serves as the sole front-end interface for system-user interaction, responsible for the initiation and final delivery of the entire process. It utilizes responsive front-end frameworks such as Vue / React to ensure a smooth experience across different terminal devices and leverages WebGL technology for high-quality visual rendering. Its core functions cover the entire process: initially receiving user questions and displaying function options; guiding users during the process and collecting key information such as their name, ID photo, voice sample, and corresponding text; and finally, providing feedback to the user and displaying the download interface for the digital avatar audio.

[0075] The information verification module is the first line of defense ensuring the accuracy and security of the digital avatar. It is specifically responsible for performing multi-dimensional, automated compliance checks on user-submitted information. It integrates several advanced algorithms: using OCR recognition and a face liveness detection model (MTCNN + Triplet Loss) to verify the authenticity of the passport photo; using a voice feature extraction algorithm (MFCC) to verify the integrity of the voice sample; and employing a text matching model (CTC) to verify the consistency between the voice and text content. Only information that passes all verifications is deemed "compliant" and proceeds to subsequent modules.

[0076] The UI dynamic update module is responsible for the intelligent evolution of the user interface. Based on the results of the information verification module, it dynamically expands UI options. For example, after successful verification, it automatically adds advanced features such as "clone by ID card" to the user. Its technical implementation relies on a component-based UI architecture. The backend pushes update commands to the frontend via a WebSocket interface, and the frontend uses Vue's dynamic rendering directives (such as v-if / v-for) to achieve real-time updates and rendering of components. The entire process requires no page refresh, providing users with a smooth and seamless interactive experience.

[0077] The request processing module serves as the central hub connecting front-end interaction and back-end generation. It is responsible for formatting and saving user clone requests, then using a message queue mechanism (such as RabbitMQ) to have AI agents automatically retrieve the requests and distribute them to appropriate digital clone agents for processing, thus achieving automated request flow. During this process, the module also employs structured data parsing algorithms to ensure accurate information interpretation and utilizes hash verification mechanisms to guarantee data integrity.

[0078] The digital avatar model generation and feedback module, based on the compliance information processed by the aforementioned modules, performs the final synthesis of the digital avatar. It uses the VITS TTS speech synthesis model to generate high-fidelity digital avatar audio and leverages a 3D model lip-sync algorithm to match the lip movements of the digital avatar with the audio timing in real time, generating a realistic audio-visual synchronization effect. Finally, the results are securely delivered to the user via an AES-256 encrypted download link.

[0079] In summary, these five modules form an interconnected, data-driven integrated system. The user interaction module is the entry point, the information verification module is the quality control station, the UI dynamic update module is the intelligent navigation, the request processing module is the scheduling center, and the digital avatar model generation and feedback module is the product terminal. Through close collaboration, they efficiently and securely transform raw user input into a personalized digital avatar.

[0080] In summary, the beneficial effects of the embodiments of this application are as follows: In traditional systems, users must manually search for and select functions within a fixed interface, a cumbersome and time-consuming process. This invention introduces a dynamic UI update mechanism, automatically expanding the corresponding function options after user information verification, transforming the previously user-driven exploration path into a system-guided one. Real-world testing data shows that the average user operation time is significantly reduced from 30 seconds in existing technologies to 5 seconds, improving interaction efficiency by up to 83% and greatly optimizing the user experience.

[0081] Existing technologies often suffer from insufficient information verification, leading to issues such as audio-visual asynchrony and mismatched identity characteristics in the generated digital clones. This invention implements a multi-dimensional information verification mechanism to rigorously check the authenticity of ID photos and the consistency between audio and text, fundamentally improving the quality of input data. As a result, the audio-text matching error rate for digital clones is reduced from 25% in existing technologies to only 2%, while the consistency between image and real identity is increased from 70% to 95%, ensuring highly realistic and reliable cloning results.

[0082] To address the response delays and processing bottlenecks caused by reliance on manual intervention in existing technologies, this invention constructs a fully automated processing chain of "AI agent - digital clone agent". This design achieves seamless integration from request distribution to parameter parsing, drastically reducing request processing latency from 5 minutes to 10 seconds. Simultaneously, the system's concurrent processing capacity has achieved a leapfrog growth, increasing from 100 times per minute to 500 times per minute, demonstrating strong potential for scalable service.

[0083] In terms of data security, this invention surpasses traditional encrypted storage methods, innovatively employing a dual security mechanism of "RSA asymmetric encryption + blockchain notarization." This solution not only controls the risk of personal information leakage to below 0.1%, ensuring compliance with the stringent management requirements for sensitive information under laws such as the Personal Information Protection Law, but also, through the immutable nature of blockchain technology, solidifies user authorization records, effectively mitigating the risk of digital avatar technology being used for identity forgery and other abuses, thus establishing a reliable foundation of technological trust.

[0084] This application also provides corresponding devices and computer storage media for implementing the digital clone model generation scheme provided in this application.

[0085] The device includes a memory and a processor. The memory is used to store instructions or code, and the processor is used to execute the instructions or code to cause the device to perform the digital clone model generation method according to any embodiment of this application.

[0086] The computer storage medium stores code, and when the code is run, the device running the code implements the digital clone model generation method described in any embodiment of this application.

[0087] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems or apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple, and relevant parts can be referred to the method section.

[0088] It should be understood that in this application, "at least one" refers to one or more items, and "more" refers to two or more items. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one" or similar expressions refer to any combination of these items, including any combination of singular or plural items. For example, "at least one" of a, b, or c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.

[0089] It should be understood that the terms center, longitudinal, transverse, up, down, front, back, left, right, vertical, horizontal, top, bottom, inside, outside, etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the present invention.

[0090] It should be noted that, unless otherwise explicitly specified and limited, the terms installation, connection, and linking should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections or electrical connections; they can refer to direct connections or indirect connections through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0091] It should also be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the statement "comprising a..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0092] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0093] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for generating a digital clone model, characterized in that, include: Obtain personal information submitted by the user for generating a digital avatar, the personal information including at least one type of data used to characterize the user's voice and / or user appearance; The personal information will be verified for compliance. If the personal information passes the verification, at least one new clone function option will be generated in the user interface; In response to the user's selection of the cloning function option, the verified personal information is associated with the cloning type corresponding to the user's selected option and encapsulated into a structured cloning request. The cloning request is obtained and distributed to the corresponding digital clone agent, so that the digital clone agent can parse the key parameters required to generate the digital clone from the cloning request; Based on the key parameters, a digital avatar model matching the user's personal information is generated. The digital avatar model is a personalized digital human model generated based on the user's voice and / or image.

2. The method according to claim 1, characterized in that, Before obtaining the personal information submitted by the user for generating the digital avatar, the process also includes: In response to the user's request to generate a digital clone model, the initial function options are displayed in the user interface; Based on the initial function options selected by the user, guide information is generated and displayed on the user interface so that the user can enter personal information on the user interface based on the guide information.

3. The method according to claim 1, characterized in that, The compliance verification of personal information includes at least one of the following: The system uses OCR to identify the name in the ID photo and matches it with the name submitted by the user, and uses a face liveness detection model to determine whether the ID photo is a real face. The sound features in the sound samples are extracted using the MFCC algorithm, and the integrity of the sound features is determined. The CTC text matching model is used to compare the text identified from the sound samples with the corresponding text submitted by the user.

4. The method according to claim 1, characterized in that, The process involves obtaining the cloning request and distributing it to the corresponding digital clone agent. The digital clone agent then parses the key parameters required to generate the digital clone from the cloning request, including: The AI ​​agent listens to the message queue and captures the encapsulated clone requests; Based on the clone type encapsulated in the clone request and the load status of each digital clone agent, the clone request is distributed to the corresponding digital clone agent so that the digital clone agent runs a structured data parsing algorithm to extract the image parameters, sound parameters and text parameters from the clone request as the key parameters required to generate the digital clone.

5. The method according to claim 4, characterized in that, The process of generating a digital avatar model that matches the user's personal information based on the key parameters includes: Based on the sound parameters parsed from the cloning request, a digital clone audio that matches the user's voice timbre is generated; based on the image parameters parsed from the cloning request, a digital clone image is generated, and a lip-sync algorithm is used to match the lip movements of the digital clone image with the time frame of the digital clone audio to obtain a digital clone model.

6. The method according to claim 1, characterized in that, The step of associating the verified personal information with the clone type corresponding to the user's selected option and encapsulating it into a structured clone request includes: Sensitive data in the personal information is encrypted, and a corresponding unique request identifier is generated; The encrypted personal information, the unique identifier of the request, the clone type corresponding to the user's selected option, and the user authorization information are jointly encapsulated into a structured clone request. The user authorization information is used to characterize the scope and validity period of the user's authorization for the use of personal information.

7. The method according to claim 6, characterized in that, The method further includes: The encrypted structured request is persistently stored in a distributed database, and user authorization information, including the scope and validity period, is stored through a consortium blockchain, whose nodes include system servers and user terminals.

8. The method according to claim 1, characterized in that, The step of generating at least one new clone functionality option in the user interface includes: The backend service sends a dynamic update command to the user interface. The dynamic update command contains configuration information for the new option, so that the user interface can add the new clone function option to the user interface using a component-based rendering method based on the dynamic update command.

9. The method according to claim 1, characterized in that, The method further includes: The audio files and preview video of the digital clone model are packaged into a download package, and a download link for the download package is generated; The download link is sent to the user interface, which then displays audio download options, a preview area, and operation instructions based on the download link. The lip movements of the digital avatar shown in the preview video are synchronized with the audio content.

10. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the digital clone model generation method as described in any one of claims 1-9.

Images