Digital human rapid development method and system, electronic equipment and storage medium

Abstract

The invention discloses a digital human rapid development method and system, electronic equipment and a storage medium, and relates to the field of digital human development. In the method, when a first figure demand and a first scene demand are received, a first appearance feature and a character feature are determined according to the first figure demand, and a second appearance feature and a preset function are determined according to the first scene demand; constructing a first digital person according to the first appearance feature, the second appearance feature, the character feature and a preset function; constructing an activity scene and a test situation according to the first scene demand, wherein the test situation comprises a trigger condition, an interaction object and an expected result; and training the first digital person according to the test situation to obtain a second digital person. By implementing the technical scheme provided by the invention, the effect of quickly developing the digital human capable of dealing with complex scenes and emotional interaction is achieved.

Description

Technical Field

[0001] The present application relates to the technical field of digital human development, and in particular to a digital human rapid development method, system, electronic device and storage medium. Background Art

[0002] With the development of computer technology, digital humans, as a bridge between the virtual and real worlds, have been widely used in many fields, such as virtual customer service, virtual anchors, and personalized entertainment. Digital technology has not only improved the efficiency of content creation in the entertainment industry, but also promoted the application of artificial intelligence and virtual reality technology.

[0003] In the prior art, in order to quickly build a digital human that meets specific needs, the methods usually adopted include but are not limited to: using a ready-made 3D model library for rapid configuration, combining animation production tools to customize the appearance characteristics and action performance of the digital human; writing interactive scripts for specific scenarios according to development needs to simulate the interaction and feedback in the real environment to realize the basic functions of the digital human. Although the above methods can meet the demand for digital human development efficiency to a certain extent, they are still insufficient when dealing with situations involving complex character traits and emotional interactions. Specifically, in the traditional digital human creation process, the integration between appearance characteristics, personality characteristics and preset functions needs to be further strengthened, especially personalized preset functions are often difficult to fully adapt to the needs of specific application scenarios.

[0004] Therefore, a method is needed to develop digital humans that can cope with complex scenarios and emotional interactions. Summary of the invention

[0005] The present application provides a digital human rapid development method, system, electronic device and storage medium, which can quickly build and optimize a digital human according to needs.

[0006] In a first aspect of the present application, a digital human rapid development method is provided, which is applied to a digital human development platform. The digital human rapid development method comprises: When a first character requirement and a first scene requirement are received, a first appearance feature and a personality feature are determined according to the first character requirement, and a second appearance feature and a preset function are determined according to the first scene requirement, wherein the first appearance feature includes gender, age, face shape and skin color, the second appearance feature includes facial features, hairstyle and clothing, the personality features include humor, rigor and gentleness, and the preset functions include voice interaction, action performance and emotional communication; Constructing a first digital human according to the first appearance feature, the second appearance feature, the personality feature and the preset function; Constructing an activity scenario and a test scenario according to the first scenario requirement, wherein the test scenario includes a trigger condition, an interaction object, and an expected result; The first digital human is trained according to the test scenario to obtain a second digital human.

[0007] By adopting the above technical solution, receiving and parsing the first person needs and the first scene needs, the appearance, personality characteristics and required functions of the digital human can be quickly determined, thereby realizing efficient customized development. The first appearance characteristics (gender, age, face shape and skin color) and the second appearance characteristics (facial features, hairstyle and clothing), as well as personality characteristics (humor, rigor, gentleness) and preset functions (voice interaction, action performance, emotional communication) are determined respectively, ensuring the comprehensiveness and personalization of the digital human. Combining the first and second appearance characteristics, the constructed digital human can highly restore the real person in appearance and enhance the user's visual realism. According to the setting of personality characteristics, the digital human can show specific behavior patterns and emotional reactions, making it closer to the real person or meeting the needs of specific scenes. The preset functions include voice interaction, action performance and emotional communication. The integration of these functions enables the digital human to interact with the user more naturally and comprehensively. By constructing activity scenes and test scenarios according to scene requirements and conducting targeted training on the digital human, the adaptability and expressiveness of the digital human in different scenes are improved. By constructing test scenarios to train the digital human, potential problems can be discovered and solved in a timely manner, thereby continuously optimizing the performance and performance of the digital human.

[0008] Optionally, the training the first digital human according to the test scenario to obtain the second digital human includes: Acquire first limb motion data of a real human in the test situation, and perform a first processing on the first limb motion data to obtain second limb motion data, wherein the first processing includes removing noise, smoothing a motion curve, and adjusting a motion speed; Each joint angle and position information in the second limb motion data is mapped to a position corresponding to the first digital human to obtain a second digital human.

[0009] By adopting the above technical solution, the first limb motion data of the real human is obtained, which can ensure that the action of the digital human is based on the real human motion data, thereby increasing the authenticity and credibility of the action. The first processing of removing noise, smoothing the action curve and adjusting the action speed of the first limb motion data can effectively reduce the error and jitter in the data, making the action of the digital human smoother and more natural, in line with the law of human movement. Accurately mapping each joint angle and position information in the processed second limb motion data to the corresponding position of the first digital human can ensure that the action of the digital human is highly consistent with the action of the real human in detail. This precise mapping not only improves the accuracy of the action of the digital human, but also enhances its expressiveness and immersion. Since the method is trained based on the real human action data in the test situation, it can flexibly respond to the action requirements under different scenarios and needs. Whether it is a simple daily action or a complex professional action, it can be trained and mapped by this method. With the change of the test situation or the update of the demand, the new limb motion data can be re-acquired and processed, and the first digital human can be re-trained to obtain a second digital human adapted to the new scene or new demand. This flexibility and adjustability make the embodiment of the present application have a wide range of application prospects and the possibility of continuous optimization. Since the second digital human's movements are more natural and realistic, the user's immersion and experience quality when interacting with the digital human can be significantly improved. This improvement not only helps to enhance the user's sense of identity and trust in the digital human, but also promotes a more natural and smooth interaction process.

[0010] Optionally, the training the first digital human according to the test scenario to obtain the second digital human includes: Acquire first facial expression data of real humans under various emotions in the test situation, and perform second processing on the first facial expression data to obtain second facial expression data, wherein the second processing includes removing noise, classifying expressions according to emotion categories, defining a unified expression description language, expression coding standards, and expression evaluation standards; The current emotional state of the first digital human is identified, a matching facial expression is selected according to the current emotional state, and the matching facial expression is mapped to the first digital human to obtain a second digital human.

[0011] By adopting the above technical solution, the first facial expression data of real humans under various emotions in the test situation are obtained, providing a rich emotional expression basis for the digital human. These data ensure the authenticity and credibility of the digital human when expressing emotions. The second processing of the first facial expression data, such as removing noise and classifying expressions according to emotional categories, improves the purity and availability of the data. At the same time, the definition of a unified expression description language, expression coding standard and expression evaluation standard provides a standardized framework for the processing and application of expression data, further improving the accuracy and consistency of expressions. By identifying the current emotional state of the first digital human, the system can accurately determine the type of emotion it needs to express. This function enables the digital human to adjust its expression more intelligently according to the current situation, enhancing the authenticity and naturalness of the interaction. According to the current emotional state, matching facial expressions are selected, and these expressions are mapped to the first digital human, thereby obtaining the second digital human. This process achieves accurate matching of emotions and expressions, making the digital human more delicate and vivid when expressing emotions. Because the second digital human can express various emotions more accurately, users will feel a stronger sense of immersion and reality when interacting with it. This sense of immersion helps to improve user participation and satisfaction. Through accurate expression mapping, the second digital human can better understand the user's emotional needs and respond with corresponding expressions. This emotional communication not only enhances the interactivity between the user and the digital human, but also promotes the emotional connection and trust between the two parties. The above process involves multiple technical fields such as emotion recognition, expression classification, and expression mapping, which promotes the continuous innovation and development of these technologies. The progress of these technologies provides strong support for the further application of digital human technology.

[0012] Optionally, the training the first digital human according to the test scenario to obtain the second digital human includes: Setting the speech speed of the first digital person according to the personality characteristics, setting the timbre of the first digital person according to the age and gender, and setting the tone of the first digital person according to the current emotional state of the first digital person and the content of the conversation; The first digital person is trained for dialogue according to a preset dialogue script to obtain a second digital person.

[0013] By adopting the above technical solution, the speech speed is set according to personality characteristics, the timbre is set according to age and gender, and the tone is set according to the current emotional state and the content of the conversation. These personalized settings make the digital human more in line with human behavior habits and emotional expression methods when expressing, thereby improving the naturalness and realism of the expression. By adjusting the tone according to the current emotional state, the digital human can convey his emotions more accurately and enhance the emotional communication ability with the user. This emotional adaptability makes the digital human more vivid and interesting in the interaction process, improving the user's participation and satisfaction. Through the preset dialogue script, the first digital human after the setting is trained to be familiar with and master the expression methods in various dialogue scenarios and contexts, thereby improving the fluency and coherence of the dialogue. This training also helps the digital human better understand and respond to the user's input and achieve more efficient interaction. Digital humans with rich expressive abilities can be applied to multiple fields, such as virtual customer service, online education, entertainment interaction, etc. In these fields, digital humans can provide a more natural, smooth and personalized interactive experience to meet the diverse needs of users. By improving the expressive ability of digital humans, users can experience a more real and vivid interactive experience, thereby improving their recognition and satisfaction with digital humans. This positive user experience helps to enhance users' trust and acceptance of digital human technology. The continuous exploration and application of technologies such as the setting of digital human speech speed, timbre, intonation, and dialogue training have promoted the continuous innovation and development of digital human technology. These technological achievements not only improve the expressive ability of digital humans, but also provide valuable experience and reference for further research and application of digital human technology in the future.

[0014] Optionally, constructing the first digital human according to the first appearance feature, the second appearance feature, the personality feature and the preset function includes: constructing an initial 3D model and creating a skeletal system, the skeletal system including a plurality of joints and bones; Binding the skeleton system to the initial 3D model so that the initial 3D model can be driven to change accordingly when the skeleton moves; Adjusting the initial 3D model according to the first appearance feature, the second appearance feature, and the personality feature to obtain an initial digital human; The preset function is added to the initial digital human to construct the first digital human.

[0015] By adopting the above technical solutions, the initial 3D model is constructed and the skeleton system is created, which provides a solid foundation for the digital human. The introduction of the skeleton system makes the movements of the digital human more natural and smooth, and also facilitates the subsequent appearance adjustment. The initial 3D model is adjusted according to the first appearance features (such as face shape, hairstyle, etc.), the second appearance features (such as clothing, accessories, etc.) and personality features, so that the final initial digital human has a high degree of personalization. This personalization is not only reflected in the appearance, but also can convey the personality traits of the digital human by fine-tuning the model details. The skeleton system is bound to the initial 3D model so that the movement of the skeleton can directly drive the model to produce corresponding changes. This design greatly enhances the action performance ability of the digital human, enabling it to make various complex movements and expressions, and improves the realism and immersion of the interaction. Adding preset functions such as voice recognition, natural language processing, and emotional computing to the initial digital human makes the first digital human not only have realistic appearance and movements, but also can interact deeply with users. The integration of these functions greatly expands the application scenarios and use value of the digital human. The entire construction process adopts a modular design concept, including modules such as 3D modeling, skeleton binding, appearance adjustment, and function integration. This design allows the development team to flexibly adjust the development process according to actual needs and improve development efficiency. At the same time, the modular design also helps to reduce development costs because developers can reuse existing modules to build new digital people.

[0016] Optionally, constructing an activity scenario and a test scenario according to the first scenario requirement includes: When the activity scenario is an interview, the interviewer is constructed as an interactive object in the test scenario, the interviewer's self-introduction is constructed as a trigger condition, and the preset questions and preset answers are determined according to the self-introduction; A first emotional state is determined according to the similarity between the interviewee's answer and the preset answer, a second emotional state is determined according to the emotion of the interviewee's answer, and a reply state of the first digital person is determined according to the first emotional state and the second emotional state.

[0017] By adopting the above technical solution, the first digital human can intelligently generate responses by presetting questions and answers, and automatically judging the emotional state and answer quality of the interviewee according to the content of the interviewee's answer. This intelligent question-and-answer method not only improves the efficiency of the interview, but also makes the interview process more objective and accurate. According to the interviewee's emotional state and answer content, the first digital human can adjust its response status to provide more personalized feedback. This personalized feedback helps to establish a better interview atmosphere and enhance the interviewee's sense of participation and trust. By constructing interview activity scenarios and test situations to simulate a real interview environment, the interviewee can demonstrate his or her abilities and qualities in a situation close to the real thing. This real scene simulation helps to improve the credibility of the interview results. The interviewee's answer content and emotional state will affect the first digital human's response status, forming a dynamic interactive process. This dynamic interaction not only makes the interview process more vivid and interesting, but also helps to tap the interviewee's deeper abilities and qualities. By presetting questions and answers and an automated emotional judgment mechanism, the first digital human can quickly process the interviewee's answers and generate corresponding responses. This automated processing method greatly improves the efficiency of the interview. Based on the comprehensive analysis of the interviewee's answers and emotional state, First Digital Human can more accurately assess the interviewee's ability and quality. This accurate assessment helps companies select talents more scientifically. First Digital Human can serve as an interviewer for remote interviews, enabling interviews across geographical restrictions. This remote interview method not only saves time and costs for companies and interviewers, but also improves the flexibility and convenience of interviews. By analyzing the interviewee's performance and feedback, First Digital Human can also provide interviewers with personalized training suggestions and guidance. This personalized training helps improve the interviewee's ability and quality and cultivate more outstanding talents for the company.

[0018] Optionally, determining the first emotional state according to the similarity between the interviewee's answer and the preset answer, determining the second emotional state according to the interviewee's answer emotion, and determining the reply state of the first digital person according to the first emotional state and the second emotional state includes: Using natural language processing technology to calculate the similarity between the answer content and the preset answer, matching the first emotional state according to the similarity, using sentiment analysis technology to analyze the emotional words in the answer content to determine the answer emotion of the interviewee, and matching the second emotional state according to the answer emotion; A corresponding target state is matched from a preset emotion table according to the first emotion state and the second emotion state, and the target state is used as a reply state of the first digital person.

[0019] By adopting the above technical solution, by calculating the similarity between the answer content and the preset answer, the closeness of the interviewee's answer to the standard answer can be objectively evaluated. This text-based similarity calculation reduces the subjectivity of human judgment and improves the accuracy of the evaluation. Sentiment analysis technology can identify emotional words in the answer content and judge the interviewee's answer emotion based on this. This technology makes the evaluation of the interviewee's emotional state more detailed and accurate, which helps to understand the interviewee's psychological state more comprehensively. According to the first emotional state (based on the similarity between the answer content and the preset answer) and the second emotional state (based on the answer emotion), the corresponding target state is matched from the preset emotional table. This process ensures that the first digital person's response can be personalized according to the specific performance and emotional state of the interviewee. This targeted response helps to establish a more positive and effective communication atmosphere. Since the first digital person's response state is dynamically adjusted according to the interviewee's real-time performance and emotional state, it can better adapt to changes in the interview process and improve the flexibility and adaptability of the response. The entire evaluation and response process is automated, reducing the need for manual intervention and improving the efficiency of the interview. At the same time, automated processing also reduces the possibility of human error and improves the reliability of the interview. Through personalized responses and targeted evaluations, First Digital Human can provide interviewees with a more intimate and professional interview experience. This experience helps to enhance the interviewee's sense of participation and satisfaction and improve the overall interview effect. This process combines natural language processing technology and sentiment analysis technology, demonstrating the integrated application of different technical fields. This technical integration not only enriches the means and methods of interview evaluation, but also provides useful reference and inspiration for technological innovation in other fields. Applying natural language processing technology and sentiment analysis technology to interview scenarios is an innovative application attempt. This innovative application not only promotes the development and progress of related technologies, but also brings new opportunities and challenges to the interview field.

[0020] In a second aspect of the present application, a digital human rapid development system is provided, including a feature module, a construction module, a scenario module and a training module, wherein: a feature module configured to, when receiving a first character requirement and a first scene requirement, determine a first appearance feature and a personality feature according to the first character requirement, and determine a second appearance feature and a preset function according to the first scene requirement, wherein the first appearance feature includes gender, age, face shape and skin color, the second appearance feature includes facial features, hairstyle and clothing, the personality features include humor, rigor and gentleness, and the preset functions include voice interaction, action performance and emotional communication; A construction module, configured to construct a first digital human according to the first appearance feature, the second appearance feature, the personality feature and the preset function; A scenario module, configured to construct an activity scenario and a test scenario according to the first scenario requirement, wherein the test scenario includes a trigger condition, an interactive object, and an expected result; A training module is configured to train the first digital human according to the test scenario to obtain a second digital human.

[0021] In the third aspect of the present application, an electronic device is provided, including a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes any one of the methods described above.

[0022] In a fourth aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores instructions, and when the instructions are executed, any of the methods described above is executed.

[0023] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. Able to quickly generate digital humans that meet the requirements based on the character and scene requirements proposed by users. By clearly defining the first appearance features, second appearance features, personality features and preset functions, the development process of digital humans is made more efficient; allowing users to customize the gender, age, face shape, skin color, facial features, hairstyle, clothing and personality features of digital humans as needed, ensuring that digital humans can accurately match the requirements of specific scenes and roles; 2. Digital humans not only have basic appearance features, but also integrate preset functions such as voice interaction, action performance and emotional communication. These functions enable digital humans to interact with users more naturally and richly in actual applications; by building activity scenarios and test situations, digital humans are trained so that they can show corresponding behaviors and reactions in different scenarios, which enhances the adaptability and flexibility of digital humans; 3. Through the refined setting of appearance and personality characteristics, as well as rich preset functions, the digital human is closer to the real person in appearance and behavior, thus enhancing the user's sense of immersion and reality; the integration of functions such as voice interaction, action performance and emotional communication enables the digital human to interact with the user in a more in-depth and meaningful way, enhancing the richness and fun of the user experience; 4. By building and training the automated process of digital humans, the degree of human intervention is reduced, thereby improving development efficiency and reducing development costs; when the digital human is developed and trained, its model and parameters can be reused in other similar projects or scenarios, further improving development efficiency and resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flowchart of the digital human rapid development method disclosed in the embodiment of the present application; Figure 2 It is a module diagram of the digital human rapid development system disclosed in the embodiment of the present application; Figure 3 It is a structural schematic diagram of an electronic device disclosed in an embodiment of the present application.

[0025] Explanation of the accompanying drawings: 201, feature module; 202, construction module; 203, scenario module; 204, training module; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory. DETAILED DESCRIPTION

[0026] In order to enable technicians in this field to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0027] In the description of the embodiments of the present application, words such as "for example" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "for example" or "for example" is intended to present related concepts in a specific way.

[0028] In the description of the embodiments of the present application, the meaning of the term "multiple" refers to two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "include", "comprise", "have" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0029] This embodiment discloses a digital human rapid development method, which is applied to a digital human development platform. Figure 1 is a flowchart of the digital human rapid development method disclosed in the embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps: S110, when a first character requirement and a first scene requirement are received, determining a first appearance feature and a personality feature according to the first character requirement, and determining a second appearance feature and a preset function according to the first scene requirement, wherein the first appearance feature includes gender, age, face shape, and skin color, the second appearance feature includes facial features, hairstyle, and clothing, the personality features include humor, rigor, and gentleness, and the preset functions include voice interaction, action performance, and emotional communication; The digital human development platform includes multiple models, and the following steps can be implemented through different models. The first character requirement refers to the setting requirements for the basic attributes and personality of the character, which usually comes from the storyline, game settings or user needs. According to the character requirements, the basic appearance of the character is first determined. This includes: Gender: Whether the character is male or female.

[0030] Age: The age group of the character, such as Child, Teenager, Middle-aged, or Elderly.

[0031] Face shape: the facial contour of the character, such as round face, square face, oval face, etc.

[0032] Skin Tone: The color of a character's skin, which can range from the natural range of skin tones to special colors such as non-human skin tones in sci-fi or fantasy works.

[0033] Personality traits: further characterize the character's personality traits based on the character's needs. These traits will affect the character's behavior and language style, including: Humor: Does the character have a sense of humor and can express ideas or handle situations in a lighthearted and humorous way?

[0034] Rigorousness: whether the character pays attention to details, acts meticulously, and thinks carefully.

[0035] Tenderness: Is the character considerate, caring, and compassionate toward others?

[0036] The first scene requirement refers to the character's environment, tasks, and interaction requirements with other elements (such as other characters, items, and events). Based on the scene requirements, the character's appearance is further refined and adjusted to better integrate into the environment or express the character's status. Including: Facial features: such as eye size, eyebrow shape, lip thickness, etc., which can reflect the character's emotional state or racial characteristics.

[0037] Hairstyle: Hairstyle designed according to the historical background, cultural customs or the character's personal preferences.

[0038] Costume: The character's clothing style not only reflects the character's identity and status, but is also affected by the scene atmosphere and seasonal changes.

[0039] Preset functions: To meet the needs of the scene, specific functions or abilities are designed for the character. These functions can be: Voice interaction: Characters can communicate with users or other characters through voice, increasing immersion and interactivity.

[0040] Action performance: Depending on the scene requirements, the character can perform specific actions such as walking, running, jumping, fighting, etc.

[0041] Emotional communication: Use facial expressions, body language, and voice changes to show the character’s emotional state, thereby enhancing the character’s realism and appeal.

[0042] S120, constructing a first digital human according to the first appearance feature, the second appearance feature, the personality feature, and the preset function; The basic appearance framework of the digital human is determined according to the first appearance features (gender, age, face shape and skin color). This is the most intuitive and basic identity of the digital human. The second appearance features (facial features, hairstyle and clothing) are integrated into it to further refine and enrich the appearance of the digital human. The adjustment of facial features can make it more in line with a specific race, cultural background or personality characteristics; the choice of hairstyle and the matching of clothing can show the identity, professional characteristics or personal style of the character. Personality characteristics are the manifestation of the digital human's inherent characteristics, which determine the digital human's behavioral reactions and language expressions in different situations. In the construction process, personality characteristics such as humor, rigor, and gentleness need to be given to the digital human to give it a unique personality charm. This is usually achieved through programming, so that the digital human can respond accordingly according to its personality characteristics when interacting with users or other characters. Preset functions are the basis for the digital human to realize its functions and roles. According to the scene requirements, the digital human is configured with preset functions such as voice interaction, action performance and emotional communication. The realization of these functions requires the help of advanced technical means, such as speech recognition and synthesis technology, animation technology, emotional computing technology, etc. Voice interaction: Through voice recognition technology, the digital human can understand the user's voice commands or questions, and generate corresponding answers or feedback through voice synthesis technology. Action performance: Using animation technology, a rich action library is designed for the digital human, including walking, running, jumping, fighting and other actions, as well as subtle changes in facial expressions to show its emotional state and personality characteristics. Emotional communication: Through emotional computing technology, the emotional state of the user or other characters is analyzed, and the digital human's expression, tone and behavior are adjusted accordingly to achieve more natural and real emotional communication. After completing the above steps, the first digital human needs to be comprehensively debugged and optimized. This includes checking whether the appearance features are coordinated and unified, whether the personality features are distinct and prominent, and whether the preset functions are stable and reliable. Through continuous debugging and optimization, ensure that the digital human can perfectly present the designer's intentions and expected effects.

[0043] Optionally, constructing the first digital human according to the first appearance feature, the second appearance feature, the personality feature and the preset function includes: constructing an initial 3D model and creating a skeletal system, the skeletal system including a plurality of joints and bones; Binding the skeleton system to the initial 3D model so that the initial 3D model can be driven to change accordingly when the skeleton moves; Adjusting the initial 3D model according to the first appearance feature, the second appearance feature, and the personality feature to obtain an initial digital human; The preset function is added to the initial digital human to construct the first digital human.

[0044] Building an initial 3D model is the first step in creating a digital human, which involves using 3D modeling software (such as Maya, Blender, etc.) to shape a basic 3D model. This model is usually a rough outline of the digital human and may contain basic body structures, but it has not yet been refined. The initial 3D model provides a basic framework for subsequent detail addition and function implementation. The skeletal system is the core of the animation and action performance of the digital human. It consists of multiple joints and bones, simulating the skeletal structure of the real human body. In 3D modeling software, specialized bone binding tools are usually used to create this system. Each joint and bone is precisely placed to ensure that they can correctly simulate the range of motion of the human body. Binding the skeletal system to the initial 3D model is a key step, which ensures that when the bones move, the initial 3D model can change accordingly. This binding relationship is achieved through weight distribution, that is, determining the degree of association between each part of the model (such as skin, muscle) and the bones. The correctness of weight distribution directly affects the realism and smoothness of the animation. After the skeleton system is bound to the initial 3D model, the model is carefully adjusted according to the first appearance features (gender, age, face shape and skin color), the second appearance features (facial features, hairstyle and clothing) and personality features. This includes the refinement of facial features (such as the shape and position of the eyes, nose and mouth), the design of hairstyle and the addition of clothing. At the same time, the expression and posture of the model are adjusted according to the personality characteristics to make it more in line with the inherent characteristics of the character. After the initial digital human model is adjusted, preset functions need to be added to achieve its specific role. These functions may include voice interaction, action performance, emotional communication, etc. According to the design requirements, these functions are realized through programming and integration of related technologies (such as speech recognition and synthesis, animation engine, emotional computing, etc.). For example, a speech recognition module is configured for the digital human so that it can understand the user's voice instructions; an animation engine is used to drive the skeleton system to achieve complex movements and expressions; emotional computing technology is used to simulate the emotional response of the digital human, etc. The last step is to conduct comprehensive debugging and optimization of the first digital human that has been built. This includes checking whether the appearance of the model meets expectations, whether the animation is smooth and natural, and whether the function is stable and reliable. Through continuous testing and adjustment, we ensure that the digital human can perfectly present the designer's intentions and expected effects.

[0045] By creating an initial 3D model and equipping it with a complex skeletal system, including multiple joints and bones, the digital human can simulate the structure and movement of the real human body. The existence of the skeletal system provides the basis for the dynamic performance of the digital human, ensuring its natural and smooth performance in various movements and postures. The skeletal system is accurately bound to the initial 3D model to ensure that the model can change accordingly when the bones move. This step is the key to achieving the dynamic effect of the digital human, which enables the digital human to make various complex movements and expressions, enhancing its realism and interactivity. The initial 3D model is carefully adjusted according to the first appearance feature (gender, age, face shape and skin color) and the second appearance feature (facial features, hairstyle and clothing), so that the digital human has a unique appearance. This process reflects a high degree of personalization and customization, which can meet the needs of different scenes and roles. Through programming and design methods, personality traits such as humor, rigor, and gentleness are integrated into the digital human, giving it a unique personality charm. The integration of such personality traits not only enhances the expressiveness of the digital human, but also makes it more in line with the expectations and preferences of users. The initial digital human is equipped with preset functions such as voice interaction, motion expression and emotional communication, so that the digital human can interact with users or other characters in various forms. The addition of these functions not only enriches the application scenarios of digital humans, but also enhances the user's experience and immersion. By realizing functions such as voice interaction and motion expression, digital humans can interact with users more naturally. This improvement in interactivity not only enhances the user's sense of participation and substitution, but also makes digital humans a more intelligent and interesting interactive medium. The entire construction process involves knowledge and technical means in multiple technical fields such as 3D modeling, bone binding, animation design, speech recognition and synthesis, and emotional computing. The comprehensive application of these technologies not only reflects the results of technological innovation, but also demonstrates the potential and value of technology in practical applications. The first digital human constructed not only has a high degree of realism and dynamic expression, but also has rich functions and powerful interactivity. This makes digital humans have broad application prospects and practical value in multiple fields such as games, animation, virtual reality, and education.

[0046] S130: construct an activity scenario and a test scenario according to the first scenario requirement, wherein the test scenario includes a trigger condition, an interaction object, and an expected result; The activity scene is the specific environment in which the digital human will interact and perform. It is built based on the first scene requirements, including the physical layout of the scene, lighting conditions, background elements, other characters or objects, etc. These elements together form a complete, contextual environment that provides a stage for the digital human's behavior and performance.

[0047] When building an activity scenario, there are several aspects to consider: Realism of the scene: Ensure that the scene matches the actual or expected situation so that the digital human can interact and behave naturally in it.

[0048] Diversity of scenarios: Build multiple different scenarios to test the adaptability and flexibility of digital humans in various environments.

[0049] Scenario scalability: The scenario is designed with possible future changes in requirements in mind, making it easy to modify and expand.

[0050] A test scenario is a series of specific events or situations defined in an activity scenario to test whether the digital human's reactions and behaviors are as expected. Each test scenario contains the following three key elements: Trigger: refers to the condition or event that causes the test scenario to occur. It defines the conditions for starting the test and provides a starting point for the digital human to act. For example, the user speaks a specific voice command or the digital human detects the presence of an object.

[0051] Interaction objects: Entities that interact with the digital human in a test situation, which can be users, other digital humans, objects, or the virtual environment itself. These interaction objects define the goals that the digital human needs to interact with.

[0052] Expected result: The expected result of the digital human's behavior or performance after the test situation is over. It is used to evaluate whether the digital human has performed the correct response or operation as expected. For example, the digital human should give a corresponding answer or perform a corresponding action after receiving the user's voice command.

[0053] When constructing test scenarios, there are several things to keep in mind: Comprehensive coverage: Ensure that the test scenarios cover the various situations and behaviors that the digital human may encounter to fully evaluate its performance and reliability.

[0054] Repeatability: Test scenarios should be designed so that they can be repeated so that the test can be run multiple times at different times or under different conditions.

[0055] Measurable: Define clear expected outcomes so that the digital human’s performance can be objectively assessed.

[0056] Optionally, constructing an activity scenario and a test scenario according to the first scenario requirement includes: When the activity scenario is an interview, the interviewer is constructed as an interactive object in the test scenario, the interviewer's self-introduction is constructed as a trigger condition, and the preset questions and preset answers are determined according to the self-introduction; A first emotional state is determined according to the similarity between the interviewee's answer and the preset answer, a second emotional state is determined according to the emotion of the interviewee's answer, and a reply state of the first digital person is determined according to the first emotional state and the second emotional state.

[0057] Construct an activity scene for simulating an interview based on the requirements of the first scenario. This scene should include the physical environment required for the interview (such as the conference room, table and chair layout, lighting, etc.) as well as virtual elements (such as the digital representation of the interviewee, the display interface of the interview questions, etc.). It is important to ensure that the atmosphere and details of the scene are close to the real interview environment in order to provide a realistic interaction platform for the digital human and the interviewee. In the interview scenario, the construction of the test situation mainly revolves around the interviewee's behavior and the digital human's reaction. Specifically: Interaction Object: Interviewers are constructed as interaction objects in the test situation. They are the subjects that the digital human needs to communicate with and evaluate.

[0058] Trigger condition: The interviewee’s self-introduction is set as the trigger condition. This action marks the official start of the interview and prompts the digital human to prepare subsequent questions and evaluations based on the content of the self-introduction.

[0059] Preset questions and preset answers: Based on the interviewee's self-introduction, the digital human will pre-set a series of related questions and prepare corresponding preset answers. These questions are designed to further understand the interviewee's ability, experience and attitude, while the preset answers serve as a reference standard for evaluating the quality of the interviewee's answers.

[0060] During the interview process, the digital human will dynamically adjust its response status based on the interviewee's answer. This process involves two key emotional state judgments: First emotional state: determined based on the similarity between the interviewee's answer and the preset answer. If the interviewee's answer is highly consistent with the preset answer, it may indicate that he has a deep understanding and preparation in this field, and the digital human may show a positive emotional state (such as satisfaction, encouragement). On the contrary, if the answer is very different from the preset answer, the digital human may show a more cautious or questioning emotional state.

[0061] Second emotional state: determined based on the interviewee's emotional expression during the interview. The interviewee's tone, facial expression and other non-verbal information can also convey their emotional state. Digital humans need to have the ability to recognize emotions in order to accurately capture these emotional signals and adjust their responses accordingly. For example, if the interviewee shows nervousness or anxiety, the digital human may adopt a gentler and more soothing response.

[0062] By simulating real interview scenarios and processes, the first digital human can communicate and interact with the interviewee in real time. This interactivity not only improves the authenticity of the test, but also helps to evaluate the performance of the digital human in different situations. The first emotional state (such as whether it is accurate, comprehensive, etc.) is determined based on the similarity between the interviewee's answer content and the preset answer, and the second emotional state is determined based on the interviewee's answer emotion (such as nervousness, confidence, etc.). The combination of these two emotional states makes the evaluation more comprehensive and in-depth, which helps to more accurately understand the interviewee's psychological state and performance level. The first digital human's reply state is determined based on the first emotional state and the second emotional state. This dynamic reply mechanism enables the digital human to flexibly adjust its reply content and method according to the actual situation of the interviewee, so as to better simulate the interactive process in a real interview. By presetting questions and presetting answers, the interview time and cost can be greatly saved. At the same time, the real-time evaluation ability of the digital human also makes the evaluation process more efficient and accurate. According to the performance of different interviewees, the first digital human's reply strategy and evaluation criteria can be personalized. This personalized development helps to better meet the needs of different users and improve user satisfaction and experience quality.

[0063] Optionally, determining the first emotional state according to the similarity between the interviewee's answer and the preset answer, determining the second emotional state according to the interviewee's answer emotion, and determining the reply state of the first digital person according to the first emotional state and the second emotional state includes: Using natural language processing technology to calculate the similarity between the answer content and the preset answer, matching the first emotional state according to the similarity, using sentiment analysis technology to analyze the emotional words in the answer content to determine the answer emotion of the interviewee, and matching the second emotional state according to the answer emotion; A corresponding target state is matched from a preset emotion table according to the first emotion state and the second emotion state, and the target state is used as a reply state of the first digital person.

[0064] Use natural language processing (NLP) technology to calculate the similarity between the interviewee's answer and the preset answer. NLP is a technology in the field of computer science used to understand and process human language. In this process, NLP technology analyzes the text features of the answer content and the preset answer, such as vocabulary, grammatical structure, and semantic meaning, and then calculates the similarity score between the two. The similarity score can be derived based on a variety of algorithms, such as cosine similarity, Jaccard similarity, or deep learning-based methods. According to the calculated similarity score, match the first emotional state from a preset list of emotional states. The emotional state here may be divided based on the interval of similarity scores. For example, high similarity may represent an "accurate" or "professional" emotional state, while low similarity may represent an "ambiguous" or "uncertain" emotional state. This matching process allows the digital human to evaluate the quality or accuracy of the interviewee's answer based on how close the answer content is to the preset answer. At the same time, sentiment analysis technology is used to analyze the emotional vocabulary in the interviewee's answer content to determine the emotion of his answer. Sentiment analysis is an NLP technology used to identify emotional tendencies in text, such as positive, negative, or neutral. In an interview scenario, sentiment analysis can help the digital human understand the emotional state of the interviewee when answering, such as confidence, nervousness, hesitation, etc. By analyzing the emotional words and phrases in the answer content, sentiment analysis technology can generate a sentiment score or label to describe the interviewee's answer emotion. According to the answer emotion obtained by sentiment analysis, a second emotional state is matched from a preset emotional state list. The second emotional state here is determined based on the actual emotion of the interviewee when answering, which is different from the first emotional state (based on similarity). For example, if the sentiment analysis shows that the interviewee showed a confident emotion when answering, then the second emotional state may be "confident". According to the first emotional state and the second emotional state, the corresponding target state is matched from the preset emotional table, and the target state is used as the response state of the first digital human. The preset emotional table is a database or mapping table containing different emotional states and their corresponding response strategies. By combining the first emotional state (based on the accuracy of the answer content) and the second emotional state (based on the answer emotion), the digital human can choose the most appropriate response strategy to deal with the interviewee's answer. This strategy may include providing further clarifying questions, giving positive feedback, adjusting the difficulty of questions, etc., to ensure the smooth progress of the interview process.

[0065] Using NLP technology to calculate the similarity between the answer content and the preset answer can accurately capture the accuracy and relevance of the interviewee's answer content, thereby inferring the first emotional state (such as accurate, vague, off-topic, etc.). Sentiment analysis technology can deeply analyze the emotional vocabulary in the answer content, identify the interviewee's emotional tendencies (such as positive, negative, neutral, etc.), and provide strong support for the determination of the second emotional state. By combining the first emotional state (based on content accuracy) and the second emotional state (based on emotional tendencies), the interviewee's answer can be more comprehensively understood, so as to formulate a more flexible and appropriate response strategy. This strategy not only takes into account the accuracy of the content, but also takes into account the interviewee's emotional feelings, which helps to establish a more positive and harmonious interview atmosphere. The first digital person can dynamically adjust the response status according to the interviewee's actual answer, making the interaction process more natural and smooth. This personalized response method helps to improve the interviewee's user experience and make them feel more concerned and respected. Through the preset emotional table and matching mechanism, the first digital person's response status can be quickly determined, avoiding the tedious and time-consuming manual evaluation. This not only improves the evaluation efficiency, but also reduces the impact of human factors on the evaluation results, making the evaluation results more objective and accurate. This process combines NLP technology and sentiment analysis technology, demonstrating the application potential of artificial intelligence technology in complex scenarios. This technology integration not only promotes the development and progress of related technologies, but also provides useful exploration and reference for more intelligent applications in similar scenarios in the future.

[0066] S140: Train the first digital human according to the test scenario to obtain a second digital human.

[0067] The training objectives need to be clearly defined. For example, in the context of an interview scenario, the training objectives may include improving the first digital person's ability to accurately understand the interviewer's answers, improve their ability to recognize emotions, and improve their ability to make appropriate responses based on these abilities. In addition, it may also include optimizing the digital person's reaction speed and enhancing their fluency in natural language processing. Based on the training objectives, a specific training plan should be designed. The training plan should include the following aspects: Dataset preparation: Collect and organize a large amount of interview dialogue data based on the test scenario. This data should cover different interview scenarios, answer content, and emotional states to simulate the diversity of real interviews.

[0068] Model Adjustment: Adjust the algorithmic model behind First Digital Human based on the training objectives. This may include optimizing the natural language processing model to improve text understanding, adjusting the sentiment analysis model to more accurately identify emotional states, and modifying the response generation model to generate more appropriate responses.

[0069] Training strategy: Develop a training strategy, including the setting of parameters such as training cycle, learning rate, batch size, etc. At the same time, it is also necessary to determine how to evaluate the training effect and whether model tuning and parameter adjustment are required.

[0070] According to the training plan, the first digital human is trained. The training process may include the following steps: Data preprocessing: Preprocess the collected interview dialogue data, including noise cleaning, word segmentation, sentiment labeling, etc.

[0071] Model training: The preprocessed data is fed into the adjusted algorithm model for training. During the training process, the model will continue to learn how to better understand and analyze the interviewee's answers and how to generate appropriate responses.

[0072] Effectiveness evaluation: Regularly evaluate the effectiveness of the model during the training process. This can be done by comparing it with the data in the test situation, evaluating the performance of the model in terms of answer accuracy, emotion recognition ability, and response appropriateness.

[0073] Model tuning: Tune the model based on the evaluation results. If you find that the model performs poorly in some aspects, you can improve performance by adjusting model parameters, improving algorithms, or adding training data.

[0074] After the training is completed, the second digital human (i.e. the trained first digital human) needs to be verified and tested. This can be achieved by putting the second digital human into real or simulated interview scenarios for testing to evaluate its performance in various situations. The results of verification and testing will be used to further confirm the training effect and provide feedback for subsequent optimization. Based on the results of verification and testing, the second digital human is iteratively optimized. This may include further adjusting model parameters, improving algorithms, increasing training data, etc. Through continuous iterative optimization, the performance of the second digital human can be gradually improved, making it more adaptable to complex interview scenarios and diverse interview needs.

[0075] Optionally, the training the first digital human according to the test scenario to obtain the second digital human includes: Acquire first limb motion data of a real human in the test situation, and perform a first processing on the first limb motion data to obtain second limb motion data, wherein the first processing includes removing noise, smoothing a motion curve, and adjusting a motion speed; Each joint angle and position information in the second limb motion data is mapped to a position corresponding to the first digital human to obtain a second digital human.

[0076] Obtain the body movement data of real people in specific test situations (such as interviews). This can usually be achieved by using motion capture technology, which can record and track the movement trajectory, joint angles and positions of the human body in real time. These data are the basis for subsequent processing and mapping. The original body movement data obtained often contains noise, uneven movement curves or inconsistent movement speeds, which will directly affect the effect of mapping to the digital human. Therefore, a series of processing, namely the first processing, is required for these data to obtain the second body movement data that is cleaner, smoother and has a uniform speed.

[0077] Noise removal: Remove the noise in the data through filtering and other methods to ensure the accuracy of the action curve.

[0078] Smooth motion curve: Smooth the motion curve to eliminate jitter and mutations, making the motion look more natural.

[0079] Adjust motion speed: Adjust the overall speed of the motion as needed to ensure that the digital human’s motion speed matches the speed expected of the human.

[0080] The processed second limb motion data contains the angle and position information of each joint, which will be mapped to the corresponding position of the first digital human. This step is the core of the training process, which involves accurately converting human motion data into motion instructions that the digital human can understand and execute.

[0081] Joint mapping: Ensure that each joint in the second limb motion data matches the corresponding joint in the first digital human model.

[0082] Action synchronization: The processed action data is synchronized to the digital human model in chronological order to achieve continuous playback of actions.

[0083] Detail adjustment: Make detailed adjustments to the digital human's movements as needed, such as expression, strength, etc., to enhance the realism and expressiveness of the movements.

[0084] After the above steps, the first digital human has learned to imitate the body movements of real humans in the test situation. These movements are not only more natural and smooth, but also in line with human habits. Therefore, the first digital human at this time can be regarded as the second digital human after training and optimization.

[0085] The acquired real human limb motion data often contains noise introduced by factors such as equipment accuracy and environmental interference. Through the noise removal step in the first processing, the impact of these noises on the digital human's motion can be significantly reduced, making the digital human's motion cleaner and more accurate. Smoothing can eliminate mutations and jitters in the motion data, making the motion curve smoother and more natural. This helps to improve the fluency and viewing of the digital human's motion and avoid discomfort to the user. Adjusting the motion speed according to actual needs can make the digital human's motion more in line with the rhythm and atmosphere of a specific scene. For example, in an interview scene, appropriately slowing down the motion speed can show a more calm and professional image. Accurately mapping the angle and position information of each joint in the second limb motion data to the corresponding position of the first digital human is the key to achieving a high degree of restoration of the digital human's motion. This step ensures that the digital human's motion can accurately reflect the motion characteristics of real humans, including posture, strength, speed, etc. By adjusting the mapping parameters and algorithms, the personalized customization of the digital human's motion can also be achieved. For example, according to the preferences and needs of different users, the digital human's motion style, expression changes, etc. are adjusted to make it more in line with the user's expectations. The optimized second digital human has more natural and smooth movements, which can significantly improve the interactive experience with users. In the interview scenario, this improvement can make users feel a more real and friendly interview atmosphere. Through training and adjustment, the second digital human can better adapt to different test situations and scenario requirements. Whether it is a simple self-introduction or a complex question-and-answer session, the digital human can show appropriate movements and expressions and interact smoothly with users. The optimized digital human can be applied in more fields, such as education, entertainment, and medical care. In the field of education, digital humans can be used as virtual lecturers to teach; in the field of entertainment, digital humans can be used as virtual idols to perform; in the field of medical care, digital humans can be used as auxiliary tools for rehabilitation training. These applications will greatly enrich people's lives and work methods.

[0086] Optionally, the training the first digital human according to the test scenario to obtain the second digital human includes: Acquire first facial expression data of real humans under various emotions in the test situation, and perform second processing on the first facial expression data to obtain second facial expression data, wherein the second processing includes removing noise, classifying expressions according to emotion categories, defining a unified expression description language, expression coding standards, and expression evaluation standards; The current emotional state of the first digital human is identified, a matching facial expression is selected according to the current emotional state, and the matching facial expression is mapped to the first digital human to obtain a second digital human.

[0087] In the test scenario, we first need to collect facial expression data of real people in different emotional states. This can be achieved in many ways, such as using high-definition cameras to record the reactions of real people under various emotional stimuli, or extracting data from existing facial expression databases. These data should cover a wide range of emotional types, such as happiness, sadness, anger, surprise, fear, disgust, and neutrality. Facial expression data may contain noise introduced by factors such as lighting changes, facial occlusion, and device jitter. Through image processing techniques such as filtering and denoising algorithms, these noises can be effectively removed to improve the purity and accuracy of the data. Expression data are classified according to the emotion type to ensure that each emotion has enough samples for subsequent analysis and training. This helps to build a more refined emotion recognition model. In order to achieve cross-platform and cross-system facial expression data exchange and sharing, a unified expression description language needs to be defined. This language should contain a comprehensive description of expression features, such as eye shape, mouth corner arc, eyebrow position, etc., to facilitate understanding and conversion between different systems. Formulate expression encoding standards to convert facial expression data into standardized digital or symbolic forms for computer processing and storage. This also helps to achieve automated recognition and classification of facial expression data. Establish expression evaluation criteria to assess the authenticity and validity of facial expression data. This can be based on expert ratings, user feedback, or automatic evaluation algorithms.

[0088] The current emotional state of the first digital person is identified through emotion recognition technology (such as an emotion recognition model based on machine learning). This can be achieved by analyzing the digital person's voice, text or behavior data. For example, in a conversation scenario, the emotional state of the digital person can be judged based on characteristics such as voice intonation, vocabulary selection, and pauses. According to the identified current emotional state, a matching facial expression is selected from the processed second facial expression data. This requires establishing a mapping relationship between emotional state and facial expression to ensure that the selected expression can accurately reflect the emotional state of the digital person. The selected facial expression is mapped to the facial model of the first digital person to generate the second digital person. This involves the analysis, rendering, and animation technology of facial expression data. Through fine animation control, the delicate changes and natural transitions of facial expressions can be ensured, thereby enhancing the realism and expressiveness of the digital person.

[0089] De-noising the acquired first facial expression data can eliminate the noise introduced by factors such as acquisition equipment and environmental interference, ensure the purity of facial expression data, and lay a solid foundation for subsequent processing. Classifying expressions according to emotional categories and defining a unified expression description language, expression coding standard, and expression evaluation standard will help to achieve standardization and normalization of facial expressions and improve the accuracy and consistency of digital human facial expressions. Using advanced artificial intelligence technology to identify the current emotional state of the first digital human, ensure that the digital human can accurately perceive and reflect its own emotional changes. According to the current emotional state, a matching facial expression is selected and mapped to the first digital human to achieve accurate matching and real-time interaction between emotions and expressions. This process enables the digital human to show emotional changes more vividly and enhance the realism and immersion of the user experience. Digital humans with rich and natural facial expressions can interact with users more vividly and naturally, improving user participation and satisfaction. This technology is particularly effective in the fields of virtual customer service, online education, entertainment games, etc. Through continuous learning and optimization, digital humans can gradually master more ways and skills of emotional expression and improve their intelligence and adaptability. This intelligence is not only reflected in the precise expression of facial expressions, but also in the keen perception and timely response to the user's emotional changes. This training process involves technologies in multiple fields such as facial expression recognition, emotional computing, and digital human modeling, which has accumulated valuable experience for the development of digital human technology. At the same time, through continuous technological innovation and optimization, the application and expansion of digital human technology in more fields can be promoted. Digital humans with rich emotional expression capabilities can be applied in more fields, such as virtual idols, virtual anchors, virtual customer service, etc. These applications can not only bring more vivid and interesting experiences to users, but also create more commercial value and social value for enterprises.

[0090] Optionally, the training the first digital human according to the test scenario to obtain the second digital human includes: Setting the speech speed of the first digital person according to the personality characteristics, setting the timbre of the first digital person according to the age and gender, and setting the tone of the first digital person according to the current emotional state of the first digital person and the content of the conversation; The first digital person is trained for dialogue according to a preset dialogue script to obtain a second digital person.

[0091] Adjust the speaking speed of the first digital person according to his / her personality characteristics. For example, an introverted digital person may speak slowly, while an extroverted digital person may speak quickly. This setting helps to enhance the expressiveness of the digital person's personality. Set the timbre of the digital person according to his / her age and gender. Different ages and genders often have different voice characteristics, such as children's voices are crisper, while the voices of the elderly may be deeper; men's voices are usually rougher, while women's voices are softer. Through reasonable timbre settings, the voice of the digital person can be closer to his / her identity characteristics. Adjust the tone of the first digital person according to his / her current emotional state and the content of the conversation. Tone is one of the important means of expressing emotions. By changing the pitch, speed, strength, etc. of the tone, different emotional colors and tone attitudes can be conveyed. For example, when expressing happiness, the tone may rise; when expressing sadness, the tone may sink. In order to train the digital person's conversation ability, a series of preset conversation scripts need to be prepared. These scripts should cover a variety of scenarios and topics to comprehensively test the digital person's conversation ability and adaptability. At the same time, the script should also contain rich emotional expressions and context changes to simulate the complex situations in real conversations. During the training process, the first digital person is placed in the preset conversation script, and the settings of its speech speed, timbre, intonation, etc. are checked through simulated conversations. At the same time, real-time adjustments and optimizations are made according to the performance of the digital person in the conversation to ensure that it can accurately and fluently express the content of the conversation and show an expression style that is consistent with its personality characteristics and emotional state. After the training, the conversation effect of the second digital person needs to be evaluated. The evaluation content can include the fluency, naturalness, accuracy of the conversation, and the richness and accuracy of emotional expression. The evaluation results are used to judge whether the training has achieved the expected goals, and the training plan is further adjusted and optimized accordingly.

[0092] The speed and timbre of the first digital person are set according to their personality, age and gender, so that the digital person's voice expression is more consistent with their set identity characteristics. This personalized setting helps to enhance the digital person's sense of reality and credibility, making it easier for users to resonate and connect emotionally. The tone of the first digital person is adjusted according to the current emotional state and the content of the conversation, making the digital person's voice expression more vivid and natural. This fusion of emotion and tone can convey richer emotional information and enhance the digital person's emotional expression ability. By training the first digital person after setting up through preset conversation scripts, real conversation scenarios can be simulated to help the digital person become familiar with and master various conversation skills and expressions. This training method helps to improve the naturalness and fluency of the digital person in actual conversations, so that it can interact with users more naturally. As the training progresses, the digital person's conversation ability will continue to be optimized and improved. It can understand the user's intentions and needs more accurately and give more appropriate and useful responses. This continuous optimization will enable the digital person to play its due role and value in more scenarios. Digital people with rich and natural expression capabilities can be applied to a variety of scenarios, such as virtual customer service, online education, entertainment games, etc. In these scenarios, digital humans can play different roles, interact and communicate with users, and bring users a more vivid and interesting experience. Through the process of multi-dimensional setting and training of the first digital human, a large amount of technical experience and data resources can be accumulated. These resources and experience will provide strong support for the subsequent innovation and development of digital human technology. With the continuous innovation and development of digital human technology, its application areas will continue to expand. In the future, digital human technology is expected to be applied and promoted in more fields, bringing more convenience and value to society.

[0093] This embodiment also discloses a digital human rapid development system. Figure 2 is a schematic diagram of the modules of the digital human rapid development system disclosed in the embodiment of the present application, such as Figure 2 As shown, the system includes a feature module 201, a construction module 202, a scenario module 203 and a training module 204, wherein: The feature module 201 is configured to, when receiving a first character requirement and a first scene requirement, determine a first appearance feature and a personality feature according to the first character requirement, and determine a second appearance feature and a preset function according to the first scene requirement, wherein the first appearance feature includes gender, age, face shape and skin color, the second appearance feature includes facial features, hairstyle and clothing, the personality features include humor, rigor and gentleness, and the preset functions include voice interaction, action performance and emotional communication; A construction module 202, configured to construct a first digital human according to the first appearance feature, the second appearance feature, the personality feature and the preset function; A scenario module 203, configured to construct an activity scenario and a test scenario according to the first scenario requirement, wherein the test scenario includes a trigger condition, an interactive object, and an expected result; The training module 204 is configured to train the first digital human according to the test scenario to obtain a second digital human.

[0094] Optionally, the training module 204 is configured to: Acquire first limb motion data of a real human in the test situation, and perform a first processing on the first limb motion data to obtain second limb motion data, wherein the first processing includes removing noise, smoothing a motion curve, and adjusting a motion speed; Each joint angle and position information in the second limb motion data is mapped to a position corresponding to the first digital human to obtain a second digital human.

[0095] Optionally, the training module 204 is configured to: Acquire first facial expression data of real humans under various emotions in the test situation, and perform second processing on the first facial expression data to obtain second facial expression data, wherein the second processing includes removing noise, classifying expressions according to emotion categories, defining a unified expression description language, expression coding standards, and expression evaluation standards; The current emotional state of the first digital human is identified, a matching facial expression is selected according to the current emotional state, and the matching facial expression is mapped to the first digital human to obtain a second digital human.

[0096] Optionally, the training module 204 is configured to: Setting the speech speed of the first digital person according to the personality characteristics, setting the timbre of the first digital person according to the age and gender, and setting the tone of the first digital person according to the current emotional state of the first digital person and the content of the conversation; The first digital person is trained for dialogue according to a preset dialogue script to obtain a second digital person.

[0097] Optionally, the construction module 202 is configured to: constructing an initial 3D model and creating a skeletal system, the skeletal system including a plurality of joints and bones; Binding the skeleton system to the initial 3D model so that the initial 3D model can be driven to change accordingly when the skeleton moves; Adjusting the initial 3D model according to the first appearance feature, the second appearance feature, and the personality feature to obtain an initial digital human; The preset function is added to the initial digital human to construct the first digital human.

[0098] Optionally, the scenario module 203 is configured to: When the activity scenario is an interview, the interviewer is constructed as an interactive object in the test scenario, the interviewer's self-introduction is constructed as a trigger condition, and the preset questions and preset answers are determined according to the self-introduction; A first emotional state is determined according to the similarity between the interviewee's answer and the preset answer, a second emotional state is determined according to the emotion of the interviewee's answer, and a reply state of the first digital person is determined according to the first emotional state and the second emotional state.

[0099] Optionally, the scenario module 203 is configured to: Using natural language processing technology to calculate the similarity between the answer content and the preset answer, matching the first emotional state according to the similarity, using sentiment analysis technology to analyze the emotional words in the answer content to determine the answer emotion of the interviewee, and matching the second emotional state according to the answer emotion; A corresponding target state is matched from a preset emotion table according to the first emotion state and the second emotion state, and the target state is used as a reply state of the first digital person.

[0100] It should be noted that: when the device provided in the above embodiment realizes its function, only the division of the above functional modules is used as an example. In actual application, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0101] This embodiment also discloses an electronic device, referring to Figure 3 The electronic device may include: at least one processor 301 , at least one communication bus 302 , a user interface 303 , a network interface 304 , and at least one memory 305 .

[0102] The communication bus 302 is used to realize the connection and communication between these components.

[0103] The user interface 303 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0104] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0105] Among them, the processor 301 may include one or more processing cores. The processor 301 uses various interfaces and lines to connect various parts in the entire server, and executes various functions of the server and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 305, and calling data stored in the memory 305. Optionally, the processor 301 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 301 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 301, and it can be implemented separately through a chip.

[0106] Among them, the memory 305 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 305 includes a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 305 may also be optionally at least one storage device located away from the aforementioned processor 301. As Figure 3 As shown, the memory 305 as a computer storage medium may include an operating system, a network communication module, a user interface module, and an application program of the digital human rapid development method.

[0107] exist Figure 3In the electronic device shown, the user interface 303 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 301 can be used to call the application program storing the digital human rapid development method in the memory 305. When executed by one or more processors 301, the electronic device executes one or more methods in the above-mentioned embodiments.

[0108] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for the present application.

[0109] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0110] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0111] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0112] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0113] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, 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, which is stored in a memory 305 and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory 305 includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk.

[0114] The above is only an exemplary embodiment of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification, those skilled in the art will easily think of other embodiments of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the technical field that are not recorded in the present disclosure. The description and examples are only regarded as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A digital human rapid development method, characterized in that: Applied to the digital human development platform, the digital human rapid development method includes: When a first character requirement and a first scene requirement are received, a first appearance feature and a personality feature are determined according to the first character requirement, and a second appearance feature and a preset function are determined according to the first scene requirement, wherein the first appearance feature includes gender, age, face shape and skin color, the second appearance feature includes facial features, hairstyle and clothing, the personality features include humor, rigor and gentleness, and the preset functions include voice interaction, action performance and emotional communication; Constructing a first digital human according to the first appearance feature, the second appearance feature, the personality feature and the preset function; Constructing an activity scenario and a test scenario according to the first scenario requirement, wherein the test scenario includes a trigger condition, an interaction object, and an expected result; The first digital human is trained according to the test scenario to obtain a second digital human.

2. The digital human rapid development method according to claim 1, characterized in that: The training of the first digital human according to the test scenario to obtain the second digital human comprises: Acquire first limb motion data of a real human in the test situation, and perform a first processing on the first limb motion data to obtain second limb motion data, wherein the first processing includes removing noise, smoothing a motion curve, and adjusting a motion speed; Each joint angle and position information in the second limb motion data is mapped to a position corresponding to the first digital human to obtain a second digital human.

3. The digital human rapid development method according to claim 1, characterized in that: The training of the first digital human according to the test scenario to obtain the second digital human comprises: Acquire first facial expression data of real humans under various emotions in the test situation, and perform second processing on the first facial expression data to obtain second facial expression data, wherein the second processing includes removing noise, classifying expressions according to emotion categories, defining a unified expression description language, expression coding standards, and expression evaluation standards; The current emotional state of the first digital human is identified, a matching facial expression is selected according to the current emotional state, and the matching facial expression is mapped to the first digital human to obtain a second digital human.

4. The digital human rapid development method according to claim 1, characterized in that: The training of the first digital human according to the test scenario to obtain the second digital human comprises: Setting the speech speed of the first digital person according to the personality characteristics, setting the timbre of the first digital person according to the age and gender, and setting the tone of the first digital person according to the current emotional state of the first digital person and the content of the conversation; The first digital person is trained for dialogue according to a preset dialogue script to obtain a second digital person.

5. The digital human rapid development method according to claim 1, characterized in that: The constructing of the first digital human according to the first appearance feature, the second appearance feature, the personality feature and the preset function comprises: constructing an initial 3D model and creating a skeletal system, the skeletal system including a plurality of joints and bones; Binding the skeleton system to the initial 3D model so that the initial 3D model can be driven to change accordingly when the skeleton moves; Adjusting the initial 3D model according to the first appearance feature, the second appearance feature, and the personality feature to obtain an initial digital human; The preset function is added to the initial digital human to construct the first digital human.

6. The digital human rapid development method according to claim 1, characterized in that: The constructing of the activity scenario and the test scenario according to the first scenario requirement includes: When the activity scenario is an interview, the interviewer is constructed as an interactive object in the test scenario, the interviewer's self-introduction is constructed as a trigger condition, and the preset questions and preset answers are determined according to the self-introduction; A first emotional state is determined according to the similarity between the interviewee's answer and the preset answer, a second emotional state is determined according to the emotion of the interviewee's answer, and a reply state of the first digital person is determined according to the first emotional state and the second emotional state.

7. The digital human rapid development method according to claim 6, characterized in that: Determining the first emotional state according to the similarity between the interviewee's answer and the preset answer, determining the second emotional state according to the interviewee's answer emotion, and determining the reply state of the first digital person according to the first emotional state and the second emotional state includes: Using natural language processing technology to calculate the similarity between the answer content and the preset answer, matching the first emotional state according to the similarity, using sentiment analysis technology to analyze the emotional words in the answer content to determine the answer emotion of the interviewee, and matching the second emotional state according to the answer emotion; A corresponding target state is matched from a preset emotion table according to the first emotion state and the second emotion state, and the target state is used as a reply state of the first digital person.

8. A digital human rapid development system, characterized in that: It includes feature module, construction module, scenario module and training module, among which: a feature module configured to, when receiving a first character requirement and a first scene requirement, determine a first appearance feature and a personality feature according to the first character requirement, and determine a second appearance feature and a preset function according to the first scene requirement, wherein the first appearance feature includes gender, age, face shape and skin color, the second appearance feature includes facial features, hairstyle and clothing, the personality features include humor, rigor and gentleness, and the preset functions include voice interaction, action performance and emotional communication; A construction module, configured to construct a first digital human according to the first appearance feature, the second appearance feature, the personality feature and the preset function; A scenario module, configured to construct an activity scenario and a test scenario according to the first scenario requirement, wherein the test scenario includes a trigger condition, an interactive object, and an expected result; A training module is configured to train the first digital human according to the test scenario to obtain a second digital human.

9. An electronic device, characterized in that: It includes a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed, the method according to any one of claims 1 to 7 is performed.

Images