A model abnormal switching method, device, equipment and storage medium

Abstract

The application relates to a model abnormal switching method and device, equipment and a storage medium. The method is applied to a digital population broadcast video generation node. The digital population broadcast video generation node is configured with a main digital human model and multiple backup digital human models. The method comprises the following steps: performing health detection on the main digital human model according to a preset detection period to obtain corresponding detection index values; judging whether the main digital human model is abnormal in operation according to the detection index values; if the main digital human model is abnormal in operation, a target backup digital human model is selected from the multiple backup digital human models; and the target backup digital human model is switched to a new main digital human model. It can be seen that the application can quickly identify model operation faults by periodically performing health inspection on the main digital human model. When the main digital human model is abnormal, the backup digital human model is automatically upgraded to the main digital human model to undertake business, thereby guaranteeing the continuous and uninterrupted operation of the digital population broadcast video generation service.

Description

Technical Field

[0001] This application relates to the field of intelligent digital humans, and in particular to a method, apparatus, device, and storage medium for abnormal model switching. Background Technology

[0002] With the rapid iteration of AI-powered digital human technology, digital voice-over videos have been implemented in diverse business scenarios, including short video news broadcasts, intelligent venue explanations, real-time online live streaming, and public knowledge popularization. The digital voice-over video generation module relies on digital human models to complete the entire process of real-time image rendering, audio timing alignment, and script-to-video conversion. The stable, long-term, and highly reliable operation of the digital human model is the core foundation for ensuring continuous output of digital human videos and maintaining uninterrupted business operations, directly determining the smooth operation of various online services.

[0003] Currently, most mainstream digital human voice-over video generation systems on the market adopt a single-model deployment architecture, with a single digital human model handling all voice-over video generation tasks. This architecture has the advantages of simple structure and low hardware resource overhead, and can be adapted to lightweight application scenarios with typical low concurrency and low fault tolerance requirements.

[0004] However, the single-model deployment architecture has a significant drawback: weak system fault tolerance. Once the digital human model crashes or malfunctions, it will immediately trigger a chain of problems such as interruption of the video generation process, stuttering of image rendering, and failure of final output, directly causing online business to stop and significantly restricting the overall stability and service availability of the digital human video generation service. Summary of the Invention

[0005] This application provides a model anomaly switching method, apparatus, device, and storage medium, aiming to solve the technical problem that traditional single-model deployment architectures are prone to causing interruptions in digital voice broadcast video services.

[0006] In a first aspect, embodiments of this application provide a model anomaly switching method, the method being applied to a digital human broadcast video generation node, the digital human broadcast video generation node being configured with a primary digital human model and multiple backup digital human models, the method comprising: Health checks are performed on the master digital human model according to a preset detection cycle to obtain the corresponding detection index values; Based on the detection index values, determine whether the main digital human model has experienced operational abnormalities; If the master digital human model malfunctions, a target backup digital human model shall be selected from the plurality of backup digital human models. The target backup digital human model is switched to the new master digital human model.

[0007] Optionally, the detection index value includes: interface connectivity. The step of determining whether the master digital human model has experienced operational abnormalities based on the detection index value includes: If the interface connectivity corresponding to the main digital human model is disconnected, it is determined that the main digital human model has experienced an operational abnormality. If the interface corresponding to the main digital human model is in a connected state, then the main digital human model is determined to be operating normally.

[0008] Optionally, the detection index value includes: task submission success rate. The step of determining whether the master digital human model has experienced operational abnormalities based on the detection index value includes: If the task submission success rate corresponding to the main digital human model is less than the first preset threshold, it is determined that the main digital human model has an operational abnormality. If the task submission success rate corresponding to the master digital human model is greater than or equal to the first preset threshold, then the master digital human model is determined to be operating normally.

[0009] Optionally, the detection index value includes: average generation time, and the step of determining whether the master digital human model has experienced operational abnormalities based on the detection index value includes: If the average generation time corresponding to the main digital human model is greater than the second preset threshold, it is determined that the main digital human model has an operational abnormality. If the average generation time of the main digital human model is less than or equal to the second preset threshold, the main digital human model is determined to be operating normally.

[0010] Optionally, the detection index value includes: error rate, and the step of determining whether the master digital human model has experienced operational abnormalities based on the detection index value includes: If the error rate of the master digital human model is greater than the third preset threshold, it is determined that the master digital human model has malfunctioned. If the error rate of the master digital human model is less than or equal to the third preset threshold, the master digital human model is determined to be operating normally.

[0011] Optionally, the detection index value includes: rate limiting status. The step of determining whether the master digital human model has experienced operational abnormalities based on the detection index value includes: If the rate limiting status corresponding to the main digital human model indicates that the server corresponding to the main digital human model is in a busy access state, then it is determined that the main digital human model has experienced an operational abnormality. If the rate limiting status corresponding to the main digital human model indicates that the server corresponding to the main digital human model is not in a busy access state, then the main digital human model is determined to be operating normally.

[0012] Optionally, the method further includes: If the detected index value returns to normal, then switch back to the main digital human model.

[0013] Secondly, embodiments of this application also provide a model anomaly switching device, which includes a unit for performing the above-described method.

[0014] Thirdly, embodiments of this application also provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0015] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the above-described method.

[0016] This application provides a method, apparatus, device, and storage medium for model anomaly switching. The method is applied to a digital human broadcast video generation node, which is configured with a primary digital human model and multiple backup digital human models. The method includes: performing health checks on the primary digital human model according to a preset detection cycle to obtain corresponding detection index values; determining whether the primary digital human model has experienced an operational anomaly based on the detection index values; if the primary digital human model has experienced an operational anomaly, selecting a target backup digital human model from the multiple backup digital human models; and switching the target backup digital human model to the new primary digital human model. Therefore, this application's technical solution, by periodically conducting health checks on the primary digital human model, can quickly identify model operational faults; when the primary digital human model experiences an anomaly, it automatically promotes a backup digital human model to the primary digital human model to take over the business, ensuring the continuous and uninterrupted operation of the digital human broadcast video generation service, effectively improving system business stability and service availability. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[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, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0020] Figure 1 A flowchart illustrating a model anomaly switching method provided in an embodiment of this application; Figure 2 A schematic block diagram of a model anomaly switching device provided in an embodiment of this application; Figure 3 A computer device provided in an embodiment of this application. Detailed Implementation

[0021] 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, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] The following disclosure provides numerous different embodiments or examples for implementing various structures of this application. To simplify the disclosure, specific examples and arrangements are described below. These are merely examples and are not intended to limit the scope of this application. Furthermore, reference numerals and / or letters may be repeated in different examples. Such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed.

[0023] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0024] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0025] It should also be further understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0026] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if [the described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [the described condition or event] is detected," or "in response to detection of [the described condition or event]."

[0027] To address the technical problem that traditional single-model deployment architectures in existing technologies can easily cause interruptions in digital population broadcast video services, this application provides a model anomaly switching device that can ensure the continuous and uninterrupted operation of digital population broadcast video generation services.

[0028] Figure 1 This is a flowchart illustrating a model anomaly switching method provided in an embodiment of this application. In one embodiment, the method is applied to a digital human broadcast video generation node, which is configured with a primary digital human model and multiple backup digital human models. The method includes steps S101-S104.

[0029] S101. Perform health checks on the main digital human model according to the preset detection cycle to obtain the corresponding detection index values.

[0030] The preset testing cycle is set based on practical experience, and this application does not impose any restrictions on it.

[0031] A digital human voice broadcast video generation node is used to generate digital human voice broadcast videos. This node is configured with one primary digital human model and multiple backup digital human models. This embodiment of the application establishes a standardized metadata structure for the digital human voice broadcast video generation node. This metadata structure includes, but is not limited to, a unique node identifier, node name, canvas display name, a list of digital human models, primary and backup digital human model configurations, general parameter configurations, digital human model-specific parameter configurations, input and output formats, functional descriptions, and status flags. Simultaneously, this embodiment of the application defines independent metadata for each connected digital human model, including but not limited to: a unique digital human model identifier, model name, display name, supplier, interface address, a list of parameter configuration items, a list of supported resolutions, a duration support range, capability tags, priority, weight, and health detection configuration. For example, for the Wan (2.2)-s2v model, resolution support is defined as 480P and 720P, with no duration extension parameter. For the Wan (2.2-Flash) model, resolution support is defined as 480P, 720P, and 1080P, with support for a 5-second duration configuration and the `prompt_extend` parameter. In addition, the parameter configuration list includes general parameters and model-specific parameters. General parameters are those supported by all digital human models, including but not limited to the digital human appearance, spoken text, voice-over timbre, and speech rate. Model-specific parameters are those unique to the corresponding model, including but not limited to model version, resolution, duration, prompt word expansion, and random seed number. It should be noted that this application embodiment defines standardized field specifications for each parameter item, including parameter name, display name, parameter type, value range, default value, interactive control type, and validation rules.

[0032] In one embodiment, this application supports online configuration, updating, canary release, addition of digital human models, modification of parameter configuration, and updating of digital human model capabilities for the metadata structure of digital human broadcast video generation nodes. Users only need to modify the corresponding metadata information to achieve dynamic updates of the front-end rendering interface without modifying the front-end code.

[0033] In one embodiment, this application specifies standardized encapsulation to interface with digital human models from different vendors, unifying request and response formats and effectively shielding the differences in interface protocols and parameters between different digital human models. A unified calling interface is exposed for the upper-layer rendering engine, including three core interfaces: task submission, status query, and result retrieval.

[0034] In one embodiment, for synchronous / asynchronous execution of digital human models, the adaptation layer uniformly encapsulates task state management logic, converting the task state of the digital human model into a standardized state. This standardized state includes, but is not limited to, queuing, generating, completed, and failed. Simultaneously, in this embodiment, the state values ​​of the digital human model are synchronized in real-time to the upper-layer rendering engine for rendering.

[0035] In one embodiment, when a user configures parameters, the rendering engine performs real-time validation on the user-configured parameters according to the parameter validation rules of the currently selected digital human model. The validation scope includes, but is not limited to, the length of the spoken text, the resolution selection, the duration range, and the range of random seed values. When the rendering engine fails the validation, it provides a real-time prompt and prohibits the submission of video generation tasks.

[0036] In one embodiment, this application automatically performs unified conversion on the format, encoding, and resolution of spoken videos generated from different digital human models, outputting standardized video files. These output video files can be directly used as input for video editing and video extension nodes, adapting to the sequential arrangement of workflows.

[0037] In one embodiment, all spoken videos generated by digital human models are saved to the tenant's media asset library. The tenant's media asset library records detailed source information for each spoken video. This source information includes, but is not limited to, the digital human model used, parameter configuration, generation time, and creator, and supports one-click secondary editing and copying.

[0038] S102. Based on the detection index values, determine whether the main digital human model has experienced operational abnormalities.

[0039] It should be noted that the main digital human model has two types of anomalies: recoverable anomalies and unrecoverable anomalies. Recoverable anomalies include, but are not limited to, network timeouts and temporary errors. For unrecoverable anomalies, this embodiment converts unrecoverable errors into standardized anomaly messages and triggers a switchover between the main and backup digital human models.

[0040] S103. If the primary digital human model malfunctions, select the target backup digital human model from among multiple backup digital human models.

[0041] Specifically, each backup digital human model is configured with a priority. In this embodiment of the application, the backup digital human model with the highest priority can be selected from multiple backup digital human models as the target backup digital human model.

[0042] S104. Switch the target backup digital human model to the new master digital human model.

[0043] Since the primary digital human model malfunctions, it is necessary to automatically switch to the target backup digital human model in a timely manner. This process is imperceptible to users, which ensures that the digital human voice video generation service will not be interrupted and also improves the user experience.

[0044] This application provides a method for switching between different models in case of anomalies. The method is applied to a digital human voice broadcast video generation node, which is configured with a primary digital human model and multiple backup digital human models. The method includes: performing health checks on the primary digital human model according to a preset detection cycle to obtain corresponding detection index values; determining whether the primary digital human model has experienced an operational anomaly based on the detection index values; if the primary digital human model has experienced an operational anomaly, selecting a target backup digital human model from the multiple backup digital human models; and switching the target backup digital human model to the new primary digital human model. Therefore, this application's technical solution, by periodically conducting health checks on the primary digital human model, can quickly identify model operational faults; when the primary digital human model experiences an anomaly, it automatically promotes a backup digital human model to the primary digital human model to take over the business, ensuring the continuous and uninterrupted operation of the digital human voice broadcast video generation service, effectively improving system business stability and service availability.

[0045] In one embodiment, the detection index value includes: interface connectivity. The above-mentioned S102 "determine whether the main digital human model has run abnormally based on the detection index value" specifically includes the following steps: S1021-S1022.

[0046] S1021. If the interface connectivity corresponding to the main digital human model is disconnected, it is determined that the main digital human model has encountered an operational abnormality.

[0047] S1022. If the interface corresponding to the main digital human model is in a connected state, then the main digital human model is determined to be running normally.

[0048] It should be noted that S1021-S1022 will be explained in detail below.

[0049] In this embodiment, the connectivity of the interface can be obtained through network probing techniques. For example, this can be achieved through ping probing, telnet probing, or DNS resolution verification. Alternatively, the interface response content can also be detected; for example, if the interface returns a connection timeout or connection reset, it indicates that the interface is disconnected. When the connectivity of the interface corresponding to the main digital human model is detected to be disconnected, it can be determined that the main digital human model cannot be used normally, and a digital human model switching operation needs to be performed.

[0050] In one embodiment, the detection index value includes: task submission success rate. The above-mentioned S102 "determine whether the master digital human model has run abnormally based on the detection index value" specifically includes the following steps: S1023-S1024.

[0051] S1023. If the task submission success rate corresponding to the main digital human model is less than the first preset threshold, it is determined that the main digital human model has an operational abnormality.

[0052] S1024. If the task submission success rate corresponding to the main digital human model is greater than or equal to the first preset threshold, the main digital human model is determined to be running normally.

[0053] It should be noted that S1023-S1024 will be explained in detail below.

[0054] The first preset threshold is set based on practical experience. This application does not impose any restrictions on it.

[0055] The task submission success rate is the percentage of successful uses of the master digital human model to the total number of uses of the master digital human model within a unit of time. If the task submission success rate corresponding to the master digital human model is less than a first preset threshold, the digital human model can be considered abnormal; otherwise, the master digital human model is considered to be operating normally.

[0056] In one embodiment, the detection index value includes: average generation time. The above-mentioned S102 "determine whether the main digital human model has run abnormally based on the detection index value" specifically includes the following steps: S1025-S1026.

[0057] S1025. If the average generation time of the main digital human model is greater than the second preset threshold, it is determined that the main digital human model has an abnormal operation.

[0058] S1026. If the average generation time corresponding to the main digital human model is less than or equal to the second preset threshold, the main digital human model is determined to be running normally.

[0059] It should be noted that S1025-S1026 will be explained in detail below.

[0060] The second preset threshold is set based on practical experience. This application does not impose any restrictions on it.

[0061] If the average generation time of the main digital human model is too long, exceeding the second preset threshold, and seriously affects the user experience, it can be determined that the main digital human model is abnormal, thus triggering the subsequent digital human model switching operation.

[0062] In one embodiment, the detection index value includes: error rate. The above-mentioned S102 "determine whether the master digital human model has run abnormally based on the detection index value" specifically includes the following steps: S1027-S1028.

[0063] S1027. If the error rate of the main digital human model is greater than the third preset threshold, it is determined that the main digital human model has an operational abnormality.

[0064] S1028. If the error rate of the main digital human model is less than or equal to the third preset threshold, the main digital human model is determined to be operating normally.

[0065] It should be noted that S1027-S1028 will be explained in detail below.

[0066] The second preset threshold is set based on practical experience. This application does not impose any restrictions on it.

[0067] If the main digital human model reports too many errors during use, in order not to affect the user experience, it can be determined that the main digital human model is malfunctioning, and a subsequent digital human model switching operation can be triggered.

[0068] In one embodiment, the detection index value includes: flow limiting state, and the above-mentioned S102 "determine whether the main digital human model has run abnormally according to the detection index value" specifically includes the following steps: ab.

[0069] a. If the rate limiting status corresponding to the main digital human model indicates that the server corresponding to the main digital human model is in a busy access state, then it is determined that the main digital human model has malfunctioned.

[0070] b. If the rate limiting status corresponding to the main digital human model indicates that the server corresponding to the main digital human model is not in a busy access state, then the main digital human model is determined to be running normally.

[0071] It should be noted that steps a-b will be explained in detail below.

[0072] When it is detected that the supplier's master digital human model is limiting the user's access, the master digital human model's access interface will return the rate limiting status information to indicate that the server corresponding to the current master digital human model is in a busy state. At this time, it can be directly determined that the master digital human model has malfunctioned, so as to trigger the subsequent execution of the digital human model switching operation.

[0073] In one embodiment, the method further includes: S105.

[0074] S105. If the detected index value returns to normal, switch back to the main digital human model.

[0075] Once the main digital human model is detected to have returned to normal, the engine can automatically switch back to the main digital human model based on user configuration. Alternatively, users can manually perform the switchback.

[0076] See Figure 2 , Figure 2 This is a schematic block diagram of a model anomaly switching device provided in an embodiment of this application. Corresponding to the above model anomaly switching method, this application also provides a model anomaly switching device. The model anomaly switching device includes a unit for executing the above model anomaly switching method. Specifically, the device is applied to a digital human broadcast video generation node, which is configured with a primary digital human model and multiple backup digital human models. The device includes: Execution unit 201 is used to perform health detection on the main digital human model according to a preset detection cycle to obtain the corresponding detection index value; The judgment unit 202 is used to determine whether the main digital human model has experienced an operational abnormality based on the detection index value; Selection unit 203 is used to select a target backup digital human model from the plurality of backup digital human models if the master digital human model experiences an operational abnormality. The switching unit 204 is used to switch the target backup digital human model to a new master digital human model.

[0077] In one embodiment, the detection index value includes: interface connectivity, and the judgment unit 202 is specifically used for: If the interface connectivity corresponding to the main digital human model is disconnected, it is determined that the main digital human model has experienced an operational abnormality. If the interface corresponding to the main digital human model is in a connected state, then the main digital human model is determined to be operating normally.

[0078] In one embodiment, the detection index value includes: task submission success rate, and the judgment unit 202 is specifically used for: If the task submission success rate corresponding to the main digital human model is less than the first preset threshold, it is determined that the main digital human model has an operational abnormality. If the task submission success rate corresponding to the master digital human model is greater than or equal to the first preset threshold, then the master digital human model is determined to be operating normally.

[0079] In one embodiment, the detection index value includes: average generation time, and the judgment unit 202 is specifically used for: If the average generation time corresponding to the main digital human model is greater than the second preset threshold, it is determined that the main digital human model has an operational abnormality. If the average generation time of the main digital human model is less than or equal to the second preset threshold, the main digital human model is determined to be operating normally.

[0080] In one embodiment, the detection index value includes: error rate, and the judgment unit 202 is specifically used for: If the error rate of the master digital human model is greater than the third preset threshold, it is determined that the master digital human model has malfunctioned. If the error rate of the master digital human model is less than or equal to the third preset threshold, the master digital human model is determined to be operating normally.

[0081] In one embodiment, the detection index value includes: current limiting state, and the judgment unit 202 is specifically used for: If the rate limiting status corresponding to the main digital human model indicates that the server corresponding to the main digital human model is in a busy access state, then it is determined that the main digital human model has experienced an operational abnormality. If the rate limiting status corresponding to the main digital human model indicates that the server corresponding to the main digital human model is not in a busy access state, then the main digital human model is determined to be operating normally.

[0082] In one embodiment, the switching unit 204 is further configured to: If the detected index value returns to normal, then switch back to the main digital human model.

[0083] like Figure 3 As shown, this application provides a computer device including a processor 31, a communication interface 32, a memory 33, and a communication bus 34. The processor 31, the communication interface 32, and the memory 33 communicate with each other through the communication bus 34. The memory 33 is used to store computer programs. In one embodiment of this application, when the processor 31 executes the program stored in the memory 33, it implements the model exception switching method provided in any of the foregoing method embodiments.

[0084] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program may be stored in a storage medium, which is a computer-readable storage medium. The computer program is executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0085] Therefore, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon, wherein the computer program, when executed by a processor, implements the steps of the model anomaly switching method provided in any of the foregoing method embodiments.

[0086] The storage medium is a physical, non-transient storage medium, such as a USB flash drive, external hard drive, read-only memory (ROM), magnetic disk, or optical disk, or any other physical storage medium capable of storing program code. The computer-readable storage medium can be non-volatile or volatile.

[0087] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0088] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0089] The steps in the methods of this application embodiment can be adjusted, merged, or deleted according to actual needs. The units in the apparatus of this application embodiment can be merged, divided, or deleted according to actual needs. Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0090] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0091] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0092] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Since these modifications and variations fall within the scope of the claims and their equivalents, this application also intends to include these modifications and variations.

[0093] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for abnormal model switching, characterized in that, The method is applied to a digital human broadcast video generation node, which is configured with a primary digital human model and multiple backup digital human models. The method includes: Health checks are performed on the master digital human model according to a preset detection cycle to obtain the corresponding detection index values; Based on the detection index values, determine whether the main digital human model has experienced operational abnormalities; If the master digital human model malfunctions, a target backup digital human model shall be selected from the plurality of backup digital human models. The target backup digital human model is switched to the new master digital human model.

2. The method according to claim 1, characterized in that, The detection index values ​​include: interface connectivity. The step of determining whether the main digital human model has experienced operational abnormalities based on the detection index values ​​includes: If the interface connectivity corresponding to the main digital human model is disconnected, it is determined that the main digital human model has experienced an operational abnormality. If the interface corresponding to the main digital human model is in a connected state, then the main digital human model is determined to be operating normally.

3. The method according to claim 1, characterized in that, The detection index values ​​include: task submission success rate. The step of determining whether the main digital human model has experienced operational abnormalities based on the detection index values ​​includes: If the task submission success rate corresponding to the main digital human model is less than the first preset threshold, it is determined that the main digital human model has an operational abnormality. If the task submission success rate corresponding to the master digital human model is greater than or equal to the first preset threshold, then the master digital human model is determined to be operating normally.

4. The method according to claim 1, characterized in that, The detection index values ​​include: average generation time. The step of determining whether the main digital human model has experienced operational abnormalities based on the detection index values ​​includes: If the average generation time corresponding to the main digital human model is greater than the second preset threshold, it is determined that the main digital human model has an operational abnormality. If the average generation time of the main digital human model is less than or equal to the second preset threshold, the main digital human model is determined to be operating normally.

5. The method according to claim 1, characterized in that, The detection index values ​​include: error rate. The step of determining whether the master digital human model has experienced operational abnormalities based on the detection index values ​​includes: If the error rate of the master digital human model is greater than the third preset threshold, it is determined that the master digital human model has malfunctioned. If the error rate of the master digital human model is less than or equal to the third preset threshold, the master digital human model is determined to be operating normally.

6. The method according to claim 1, characterized in that, The detection index values ​​include: rate limiting status. The step of determining whether the main digital human model is experiencing operational abnormalities based on the detection index values ​​includes: If the rate limiting status corresponding to the main digital human model indicates that the server corresponding to the main digital human model is in a busy access state, then it is determined that the main digital human model has experienced an operational abnormality. If the rate limiting status corresponding to the main digital human model indicates that the server corresponding to the main digital human model is not in a busy access state, then the main digital human model is determined to be operating normally.

7. The method according to claim 1, characterized in that, The method further includes: If the detected index value returns to normal, then switch back to the main digital human model.

8. A model anomaly switching device, characterized in that, The device is applied to a digital human broadcast video generation node, which is configured with a master digital human model and multiple backup digital human models. The device includes: An execution unit is used to perform health checks on the main digital human model according to a preset detection cycle to obtain corresponding detection index values; The judgment unit is used to determine whether the main digital human model has experienced an operational abnormality based on the detection index value. The selection unit is used to select a target backup digital human model from the plurality of backup digital human models if the master digital human model experiences an operational abnormality. The switching unit is used to switch the target backup digital human model to a new master digital human model.

9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program that, when executed by a processor, can implement the method as described in any one of claims 1 to 7.

Images