Digital human interaction system based on artificial intelligence optimization

Abstract

The invention discloses a digital human interaction system based on artificial intelligence optimization, which comprises a signal acquisition module, an emotion processing module, an image voice processing module and a feedback correction module, and is characterized in that the signal acquisition module acquires startup signal data, and voice data and text data after startup; the emotion processing module carries out word meaning interpretation on the voice data and the text data to obtain result data, and obtains the result data, corresponding emotion data, response text data and corresponding response emotion data; the image voice processing module generates an appearance image model of the digital person and a sound image model of the digital person, fuses the appearance image model and the sound image model, and generates a target digital person; and the feedback correction module is used for evaluating the feedback of the user, correcting the emotion processing module according to an evaluation result, correspondingly adjusting the emotion processing module according to the emotion change of the user, and providing a more humanized digital human dialogue service.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a digital human interaction system based on artificial intelligence optimization. Background Art

[0002] In recent years, with the rapid development of AI-generated content, the digital human industry has once again attracted widespread attention and ushered in unprecedented development opportunities. More and more digital humans have sprung up in the public eye. While becoming a key tool for companies to innovate services and improve efficiency, they are also reshaping the business ecosystem and user experience. Digital humans have played a very important role in many fields and have demonstrated great application value in promoting the digital transformation of the industry.

[0003] Currently, digital humans on the market are able to communicate with users and perform certain accompanying actions, but digital humans cannot monitor users’ emotional changes and changes in the positivity of their responses, and cannot accurately analyze users’ emotional changes and make relevant responses, making the communication between digital humans and users mechanical and fixed, and unable to meet users’ dialogue needs. Summary of the invention

[0004] The technical problem solved by the present invention is that when a digital human communicates with a user, the digital human cannot analyze the user's emotional changes and the changes in the positivity of the response based on the user's response data. There is a large difference between the emotional expression of the digital human and the emotional perception of the human senses on the overall image of the digital human.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a digital human interaction system based on artificial intelligence optimization, including a signal acquisition module, an emotion processing module, an image and speech processing module and a feedback correction module;

[0006] The signal acquisition module is used to collect power-on signal data, voice data and text data after power-on;

[0007] The emotion processing module is used to interpret the meaning of the voice data and text data, and obtain interpretation result data and emotion data corresponding to the result data, and obtain response data and response emotion data corresponding to the response text data;

[0008] The image and speech processing module generates a digital human appearance model and a digital human voice image model based on the response data and the response emotion data, and fuses the appearance model with the voice image model to generate a target digital human;

[0009] The feedback correction module is used to evaluate the user's feedback and correct the emotion processing module according to the evaluation result and the comparison result between the overall emotion data of the digital human and the response emotion data.

[0010] As a preferred solution of the digital human interaction system based on artificial intelligence optimization described in the present invention, the specific process of the signal acquisition module collecting the power-on signal data, the user voice data after power-on and the user text data is as follows:

[0011] The signal acquisition module collects the user's voice data and text data, and determines whether the user's voice data and the text data contain pre-stored keyword audio data or keyword text data:

[0012] If it contains pre-stored keyword audio data or keyword text data, it is determined that the user's voice data or the text data is power-on signal data, and the emotion processing module, the image and voice processing module, and the feedback correction module are activated, and at the same time, the signal acquisition module continues to collect the user's voice data and the user's text data, and sends the user's voice data and the user's text data to the emotion processing module;

[0013] If the pre-stored keyword audio data or keyword text data is not included, the emotion processing module, the image and speech processing module, and the feedback correction module are in a dormant state.

[0014] As a preferred solution of the digital human interaction system based on artificial intelligence optimization described in the present invention, wherein: the emotion processing module is connected to the signal acquisition module, the emotion processing module performs speech-to-text conversion on the user voice data, converts the user voice data into corresponding user text data, calls up a large language model, inputs the user text data into the large language model, the large language model interprets the meaning of the user text data, and obtains response data corresponding to the user text data, inputs the user text data and the response data into an emotion tendency model, and obtains corresponding emotion tendency data and response emotion data.

[0015] As a preferred solution of the digital human interaction system based on artificial intelligence optimization described in the present invention, wherein: the image and speech processing module includes three processing units, namely an image processing unit, a speech processing unit and an image and speech fusion processing unit;

[0016] The image processing unit generates an appearance model of the digital person based on the response data and the response emotion data;

[0017] The speech processing unit is used to generate a voice image model of a digital human based on the response data and the response emotion data;

[0018] The image-speech fusion unit is used to fuse the appearance image model and the sound image model.

[0019] As a preferred solution of the digital human interaction system based on artificial intelligence optimization described in the present invention, the specific process of the image processing unit generating the digital human appearance model based on the response data and the response emotion data is as follows:

[0020] The appearance model stores multiple sets of training data, each of which includes historical emotion data and the historical original appearance, eyes and lips of the digital human corresponding to the historical emotion data;

[0021] Each of the plurality of training data sets includes historical response data and the historical opening and closing of the lips of the digital person corresponding to the historical response data set;

[0022] Each of the plurality of training data sets includes historical response data, historical response emotion data, and historical actions of the digital human corresponding to the historical response data and the historical response emotion data;

[0023] The appearance model takes the historical emotion data of each set of training data in the multiple sets of training data as input, and outputs the historical original appearance image, eyes and lips of the digital human corresponding to the historical emotion data;

[0024] The appearance model takes the historical response data of each set of training data in the plurality of sets of training data as input, and outputs the historical opening and closing of the lips of the corresponding digital human being as output;

[0025] The appearance model takes the historical response data and historical response emotion data of each set of training data in the multiple sets of training data as input, and the historical actions of the digital human corresponding to the historical response data and the historical response emotion data as output

[0026] The response data and the response emotion data are input into the appearance model to obtain the original appearance of the digital human, the appearance of the eyes and lips, the opening and closing of the corresponding digital human lips, and the corresponding actions of the digital human.

[0027] The specific process of the voice processing unit for generating the voice image model of the digital human based on the response data and the response emotion data is as follows:

[0028] The voice image model stores multiple sets of training data, each of which includes historical emotion data and the volume and speech speed of historical voice response data corresponding to the historical emotion data. The voice image model takes the historical emotion data of each set of training data in the multiple sets of training data as input, and takes the volume and speech speed of historical voice response data corresponding to the historical emotion data as output;

[0029] Inputting the response data and the response emotion data into the voice image model to obtain the volume and speech speed of the digital human's voice response data;

[0030] The image and speech fusion processing unit fuses the output results of the image processing unit and the speech processing unit to obtain a target digital human.

[0031] The image and speech fusion processing unit fuses the output results of the image processing unit and the speech processing unit to obtain a target digital human.

[0032] As a preferred solution of the digital human interaction system based on artificial intelligence optimization described in the present invention, the training process of the emotional tendency model is as follows:

[0033] Among them, the emotional tendency model is obtained using multiple groups of training data, each group of training data includes historical text data and the pleasantness of eight major emotions and the corresponding activation degrees corresponding to the PAD emotional data model corresponding to the historical text data. By inputting the user text data and the response data into the emotional tendency model, the emotional tendency data and the response emotional data can be obtained. The emotional tendency data and the response emotional data are the pleasantness of eight major emotions and the corresponding activation degrees.

[0034] As a preferred solution of the digital human interaction system based on artificial intelligence optimization described in the present invention, the specific process of the feedback correction module evaluating the user's feedback is as follows:

[0035] The feedback correction module is connected to the image and speech fusion processing module. The feedback correction module receives and stores the user response data within a certain period of time, and inputs the user response data into the user feedback detection unit. The user feedback detection unit is used to analyze the changes in the user response data and the changes in the response positivity, and obtain the feedback result. If the user's emotion becomes negative, it is recorded as 1, and if it remains unchanged, it is recorded as 0; if the response positivity decreases, it is recorded as 1, and if it remains unchanged, it is recorded as 0. The feedback results include four types:

[0036] 0, 0 means that the emotion change remains unchanged, the response positivity remains unchanged, and no adjustment is needed;

[0037] 0, 1 means that the emotion change remains unchanged, the response enthusiasm decreases, and the activation degree of the emotion tendency model response emotion data is increased by 0.2;

[0038] 1, 0 means that the emotion change is negative, the response positivity remains unchanged, and the activation degree of the emotion tendency model response emotion data is increased by 0.4;

[0039] 1, 1 means that the emotional change becomes negative, the response enthusiasm decreases, and the emotional tendency model responds to the emotional data activation by 0.6.

[0040] As a preferred solution of the digital human interaction system based on artificial intelligence optimization described in the present invention, the specific steps of the feedback correction module correcting the emotion processing module according to the comparison result between the response data of the digital human and the response emotion data are as follows:

[0041] The input of the emotional tendency model is the response data, and the output is the response emotional data. The overall image of the target digital person is captured in real time to obtain the overall emotional data. The response emotional data is compared with the overall emotional data, and the activation degree of the emotional tendency model is adjusted until the difference in pleasure between the response emotional data and the overall emotional data is less than 0.5.

[0042] As a preferred solution of the digital human interaction system based on artificial intelligence optimization described in the present invention, wherein: the overall emotional data is the emotional data of the overall image of the target digital human evaluated by the human senses, and the overall emotional data and the response data are compared in terms of emotional pleasure;

[0043] If the pleasure of the overall emotional data is greater than the pleasure of the response data by more than 0.5, then the pleasure of the emotional tendency model response emotional data is reduced;

[0044] If the pleasure of the overall emotional data is less than the pleasure of the response data by less than 0.5, then the emotional tendency model is increased to respond to the pleasure of the emotional data.

[0045] The beneficial effects of the present invention are as follows: by collecting the conversations between the user and the digital human within a certain period of time and monitoring the changes in the length of the user's response data, the changes in the user's response enthusiasm can be reflected; by inputting the user's response data into the emotional tendency detection model, the user's emotional changes can be reflected; when the emotional changes are negative or the user's response enthusiasm decreases, the emotional activation degree in the emotional tendency monitoring model is increased accordingly, which can improve the digital human's response enthusiasm and respond to the user's emotional changes more delicately, so that the digital human can communicate with the user in a more humane way.

[0046] By inputting the response data of the digital person into the emotional tendency detection model, the response emotional data of the digital person as perceived by the computer system is obtained. The overall image of the digital person is evaluated through human senses to obtain human evaluation data. The evaluation data is input into the emotional tendency detection model to obtain the response emotional data corresponding to the human senses, which is recorded as the overall emotional data. The pleasure of the overall emotional data is compared with the pleasure of the response emotional data, and the emotional pleasure in the emotional tendency monitoring model is increased accordingly. This can make the emotions expressed by the digital person closer to the emotions felt by human senses, so that the digital person can express human-like emotions more vividly and vividly, and improve the consistency of the emotional pleasure levels of the digital person and the user. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A basic flow chart of a digital human interaction system based on artificial intelligence optimization provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0049] Example 1, reference Figure 1 , is an embodiment of the present invention, providing a digital human interaction system based on artificial intelligence optimization, including a signal acquisition module, an emotion processing module, an image and speech processing module and a feedback correction module,

[0050] The signal acquisition module is used to collect power-on signal data, voice data and text data;

[0051] The emotion processing module is used to interpret the meaning of the voice data and text data, and obtain interpretation result data and emotion data corresponding to the result data, and obtain response data and response emotion data corresponding to the response text data;

[0052] The image and speech processing module generates a digital human appearance model and a digital human voice image model based on the response data and the response emotion data, and fuses the appearance model with the voice image model to generate a target digital human;

[0053] The feedback correction module is used to evaluate the user's feedback and correct the emotion processing module according to the evaluation result and the comparison result between the overall emotion data of the digital human and the response emotion data.

[0054] The specific process of the signal acquisition module collecting power-on signal data, user voice data and user text data is as follows:

[0055] The signal acquisition module collects the user's voice data and text data, and determines whether the user's voice data and the text data contain pre-stored keyword audio data or keyword text data:

[0056] If it contains pre-stored keyword audio data or keyword text data, it is determined that the user's voice data or the text data is power-on signal data, and the emotion processing module, the image and voice processing module, and the feedback correction module are activated, and at the same time, the signal acquisition module continues to collect the user's voice data and the user's text data, and sends the user's voice data and the user's text data to the emotion processing module;

[0057] If the pre-stored keyword audio data or keyword text data is not included, the emotion processing module, the image and speech processing module, and the feedback correction module are in a dormant state.

[0058] In this embodiment, the power-on signal data is pre-stored keyword audio data or keyword text data. Once the signal acquisition module receives the power-on signal data, it activates other modules of the system and puts other modules into working state. If no power-on signal data is received, other modules are in sleep state. This operating mode can prevent the digital human from being accidentally started and reduce the energy consumption of system operation.

[0059] The emotion processing module is connected to the signal acquisition module, and the emotion processing module performs speech-to-text conversion on the user voice data, converts the user voice data into corresponding user text data, calls a large language model, inputs the user text data into the large language model, the large language model interprets the meaning of the user text data, and obtains response data corresponding to the user text data, and inputs the user text data and the response data into an emotion tendency model to obtain corresponding emotion tendency data and response emotion data.

[0060] In this embodiment, the large language model has the characteristics of large algorithm and high accuracy in interpreting natural language. By calling the large language model, the digital human can quickly interpret the user's text data and give a reasonable response data similar to human communication. Calling this model can greatly reduce the rigidity of the dialogue between the digital human and the user, allowing the digital human to communicate with the user in a more humane way.

[0061] The image and speech processing module includes three processing units, namely an image processing unit, a speech processing unit, and an image and speech fusion processing unit;

[0062] The image processing unit generates an appearance model of the digital person based on the response data and the response emotion data;

[0063] The speech processing unit is used to generate a voice image model of a digital human based on the response data and the response emotion data;

[0064] The image-speech fusion unit is used to fuse the appearance image model and the sound image model.

[0065] In this embodiment, the digital human is composed of an appearance image model and a sound image model. The digital human's image and voice are processed separately in separate units to obtain the corresponding appearance image model and sound image model. Finally, the two models are integrated to improve the processing efficiency and accuracy of the digital human's voice image.

[0066] The specific process of the image processing unit generating the digital human appearance model based on the response data and the response emotion data is as follows:

[0067] The appearance model stores multiple sets of training data, each of which includes historical emotion data and the historical original appearance, eyes and lips of the digital human corresponding to the historical emotion data;

[0068] Each of the plurality of training data sets includes historical response data and the historical opening and closing of the lips of the digital person corresponding to the historical response data set;

[0069] Each of the plurality of training data sets includes historical response data, historical response emotion data, and historical actions of the digital human corresponding to the historical response data and the historical response emotion data;

[0070] The appearance model takes the historical emotion data of each set of training data in the multiple sets of training data as input, and outputs the historical original appearance image, eyes and lips of the digital human corresponding to the historical emotion data;

[0071] The appearance model takes the historical response data of each set of training data in the plurality of sets of training data as input, and outputs the historical opening and closing of the lips of the corresponding digital human being as output;

[0072] The appearance model takes the historical response data and historical response emotion data of each set of training data in the multiple sets of training data as input, and the historical actions of the digital human corresponding to the historical response data and the historical response emotion data as output

[0073] The response data and the response emotion data are input into the appearance model to obtain the original appearance of the digital human, the appearance of the eyes and lips, the opening and closing of the corresponding digital human lips, and the corresponding actions of the digital human.

[0074] The specific process of the voice processing unit for generating the voice image model of the digital human based on the response data and the response emotion data is as follows:

[0075] The voice image model stores multiple sets of training data, each of which includes historical emotion data and the volume and speech speed of historical voice response data corresponding to the historical emotion data. The voice image model takes the historical emotion data of each set of training data in the multiple sets of training data as input, and takes the volume and speech speed of historical voice response data corresponding to the historical emotion data as output;

[0076] Inputting the response data and the response emotion data into the voice image model to obtain the volume and speech speed of the digital human's voice response data;

[0077] The image and speech fusion processing unit fuses the output results of the image processing unit and the speech processing unit to obtain a target digital human.

[0078] The training process of the emotion tendency model is as follows: wherein, the emotion tendency model is obtained using multiple groups of training data, each of the multiple groups of training data includes historical text data and a PAD emotion data model corresponding to the historical text data and corresponding eight major emotions' pleasantness and corresponding activation degrees, and the user text data and the response data are input into the emotion tendency model to obtain emotion tendency data and response emotion data, and the emotion tendency data and response emotion data are the eight major emotions' pleasantness and corresponding activation degrees.

[0079] In this embodiment, by querying the corresponding PAD emotional data, the corresponding pleasure levels of the eight emotions and the corresponding activation levels of multiple groups of historical text data, it is possible to analyze the emotions of user text data and response data in a more comprehensive and multi-dimensional manner. The eight emotions include happy, bored, dependent, contemptuous, relaxed, anxious, docile and hostile.

[0080] The specific process of the feedback correction module evaluating the user's feedback is as follows: the feedback correction module is connected to the image and speech fusion processing module, the feedback correction module receives and stores the user response data within a certain period of time, and inputs the user response data into the user feedback detection unit, the user feedback detection unit is used to analyze the emotional changes and the changes in the response positivity of the user response data to obtain the feedback result. If the user's emotion becomes negative, it is recorded as 1, and if it remains unchanged, it is recorded as 0; if the response positivity decreases, it is recorded as 1, and if it remains unchanged, it is recorded as 0. The feedback results include four types:

[0081] 0, 0 means that the emotion change remains unchanged, the response positivity remains unchanged, and no adjustment is needed;

[0082] 0, 1 means that the emotion change remains unchanged, the response enthusiasm decreases, and the activation degree of the emotion tendency model response emotion data is increased by 0.2;

[0083] 1, 0 means that the emotion change is negative, the response positivity remains unchanged, and the activation degree of the emotion tendency model response emotion data is increased by 0.4;

[0084] 1, 1 means that the emotional change becomes negative, the response enthusiasm decreases, and the emotional tendency model responds to the emotional data activation by 0.6.

[0085] In this embodiment, based on the feedback results of the user feedback detection unit, the emotional tendency model makes different adjustments for each feedback result. The digital human can output response data with a higher degree of emotional activation than before. The purpose is to make the user feel the sincerity and thoughtfulness of the other party's response when communicating with a real person. The communication between the digital human and the user is no longer just a simple mechanical communication.

[0086] The specific steps of the feedback correction module correcting the emotion processing module according to the comparison result between the response data of the digital person and the response emotion data are as follows:

[0087] The input of the emotional tendency model is the response data, and the output is the response emotional data. The overall image of the target digital person is captured in real time to obtain the overall emotional data. The response emotional data is compared with the overall emotional data, and the activation degree of the emotional tendency model is adjusted until the difference in pleasure between the response emotional data and the overall emotional data is less than 0.5.

[0088] The overall emotional data is emotional data of the overall image of the target digital person evaluated by a person's senses, and the overall emotional data is compared with the response data in terms of emotional pleasure;

[0089] If the pleasure of the overall emotional data is greater than the pleasure of the response data by more than 0.5, then the pleasure of the emotional tendency model response emotional data is reduced;

[0090] If the pleasure of the overall emotional data is less than the pleasure of the response data by less than 0.5, then the emotional tendency model is increased to respond to the pleasure of the emotional data.

[0091] By collecting conversations between users and digital humans over a certain period of time and monitoring changes in the length of user response data, we can reflect changes in the user's response enthusiasm. By inputting the user response data into the emotional tendency detection model, we can reflect the user's emotional changes. When the emotional changes are negative or the user's response enthusiasm decreases, the emotional activation in the emotional tendency monitoring model is increased accordingly, which can improve the digital human's response enthusiasm and respond to the user's emotional changes more delicately, so that the digital human can communicate with the user in a more humane way.

[0092] By inputting the response data of the digital person into the emotional tendency detection model, the response emotional data of the digital person as perceived by the computer system is obtained. The overall image of the digital person is evaluated through human senses to obtain human evaluation data. The evaluation data is input into the emotional tendency detection model to obtain the response emotional data corresponding to the human senses, which is recorded as the overall emotional data. The pleasure of the overall emotional data is compared with the pleasure of the response emotional data, and the emotional pleasure in the emotional tendency monitoring model is increased accordingly. This can make the emotions expressed by the digital person closer to the emotions felt by human senses, so that the digital person can express human-like emotions more vividly and vividly, and improve the consistency of the emotional pleasure levels of the digital person and the user.

[0093] It should be understood by those skilled in the art that the embodiments of the present invention can be provided as methods, systems or computer program products. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program codes. Among them, the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, disk or optical disk. These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0094] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A digital human interaction system based on artificial intelligence optimization, characterized by: include: Signal acquisition module, emotion processing module, image and speech processing module and feedback correction module, The signal acquisition module is used to collect power-on signal data, voice data and text data; The emotion processing module is used to interpret the meaning of the voice data and text data, and obtain result data and user emotion data corresponding to the result data, and obtain response data and response emotion data corresponding to the response data; The image and speech processing module generates a digital human appearance model and a digital human voice image model based on the response data and the response emotion data, and fuses the appearance model with the voice image model to generate a target digital human; The feedback correction module is used to evaluate the user's feedback and correct the emotion processing module according to the evaluation result and the comparison result between the overall emotion data of the digital human and the response emotion data.

2. The digital human interaction system based on artificial intelligence optimization as claimed in claim 1, characterized in that: The specific process of the signal acquisition module collecting power-on signal data, user voice data after power-on, and user text data is as follows: The signal acquisition module collects the user's voice data and text data, and determines whether the user's voice data and the text data contain pre-stored keyword audio data or keyword text data: If it contains pre-stored keyword audio data or keyword text data, it is determined that the user's voice data or the text data is power-on signal data, and the emotion processing module, the image and voice processing module, and the feedback correction module are activated, and at the same time, the signal acquisition module continues to collect the user's voice data and the user's text data, and sends the user's voice data and the user's text data to the emotion processing module; If the pre-stored keyword audio data or keyword text data is not included, the emotion processing module, the image and speech processing module, and the feedback correction module are in a dormant state.

3. The digital human interaction system based on artificial intelligence optimization as claimed in claim 2, characterized in that: The emotion processing module is connected to the signal acquisition module, and the emotion processing module performs speech-to-text conversion on the user voice data, converts the user voice data into corresponding user text data, calls a large language model, inputs the user text data into the large language model, the large language model interprets the meaning of the user text data, and obtains response data corresponding to the user text data, and inputs the user text data and the response data into an emotion tendency model to obtain corresponding emotion tendency data and response emotion data.

4. The digital human interaction system based on artificial intelligence optimization as claimed in claim 3, characterized in that: The image and speech processing module includes three processing units, namely an image processing unit, a speech processing unit, and an image and speech fusion processing unit; The image processing unit generates an appearance model of the digital person based on the response data and the response emotion data; The speech processing unit is used to generate a voice image model of a digital human based on the response data and the response emotion data; The image-speech fusion unit is used to fuse the appearance image model and the sound image model.

5. The digital human interaction system based on artificial intelligence optimization as claimed in claim 4, characterized in that: The specific process of the image processing unit generating the digital human appearance model based on the response data and the response emotion data is as follows: The appearance model stores multiple sets of training data, each of which includes historical emotion data and the historical original appearance, eyes and lips of the digital human corresponding to the historical emotion data; Each of the plurality of training data sets includes historical response data and the historical opening and closing of the lips of the digital person corresponding to the historical response data set; Each of the plurality of training data sets includes historical response data, historical response emotion data, and historical actions of the digital human corresponding to the historical response data and the historical response emotion data; The appearance model takes the historical emotion data of each set of training data in the multiple sets of training data as input, and outputs the historical original appearance image, eyes and lips of the digital human corresponding to the historical emotion data; The appearance model takes the historical response data of each set of training data in the plurality of training data as input, and outputs the historical opening and closing of the lips of the corresponding digital person; The appearance model takes the historical response data and historical response emotion data of each set of training data in the multiple sets of training data as input, and the historical actions of the digital human corresponding to the historical response data and the historical response emotion data as output; The response data and the response emotion data are input into the appearance model to obtain the original appearance of the digital human, the appearance of the eyes and lips, the opening and closing of the corresponding digital human lips, and the corresponding actions of the digital human.

6. The digital human interaction system based on artificial intelligence optimization as claimed in claim 5, characterized in that: The specific process of the voice processing unit for generating the voice image model of the digital human based on the response data and the response emotion data is as follows: The voice image model stores multiple sets of training data, each of which includes historical emotion data and the volume and speech speed of historical voice response data corresponding to the historical emotion data. The voice image model takes the historical emotion data of each set of training data in the multiple sets of training data as input, and takes the volume and speech speed of historical voice response data corresponding to the historical emotion data as output; Inputting the response data and the response emotion data into the voice image model to obtain the volume and speech speed of the digital human's voice response data; The image and speech fusion processing unit fuses the output results of the image processing unit and the speech processing unit to obtain a target digital human.

7. The digital human interaction system based on artificial intelligence optimization as claimed in claim 6, characterized in that: The training process of the sentiment tendency model is as follows: Among them, the emotional tendency model is obtained using multiple groups of training data, each group of training data includes historical text data and the pleasantness of eight major emotions and the corresponding activation degrees corresponding to the PAD emotional data model corresponding to the historical text data. The user text data and the response data are input into the emotional tendency model to obtain emotional tendency data and response emotional data, and the emotional tendency data and response emotional data are the pleasantness of eight major emotions and the corresponding activation degrees.

8. The digital human interaction system based on artificial intelligence optimization as claimed in claim 7, characterized in that: The specific process of the feedback correction module evaluating the user's feedback is as follows: The feedback correction module is connected to the image and speech fusion processing module. The feedback correction module receives and stores the user response data within a certain period of time, and inputs the user response data into the user feedback detection unit. The user feedback detection unit is used to analyze the changes in the user response data and the changes in the response positivity, and obtain the feedback results. If the user's emotion becomes negative, it is recorded as 1, and if it remains unchanged, it is recorded as 0. If the response positivity decreases, it is recorded as 1, and if it remains unchanged, it is recorded as 0. The feedback results include four types: 0, 0 means that the emotion change remains unchanged, the response positivity remains unchanged, and no adjustment is needed; 0, 1 means that the emotion change remains unchanged, the response enthusiasm decreases, and the activation degree of the emotion tendency model response emotion data is increased by 0.2; 1, 0 means that the emotion change is negative, the response positivity remains unchanged, and the activation degree of the emotion tendency model response emotion data is increased by 0.4; 1, 1 means that the emotional change becomes negative, the response enthusiasm decreases, and the emotional tendency model responds to the emotional data activation by 0.

6.

9. The digital human interaction system based on artificial intelligence optimization as claimed in claim 8, characterized in that: The specific steps of the feedback correction module correcting the emotion processing module according to the comparison result between the response data of the digital person and the response emotion data are as follows: The input of the emotional tendency model is the response data, and the output is the response emotional data. The overall image of the target digital person is captured in real time to obtain the overall emotional data. The response emotional data is compared with the overall emotional data, and the activation degree of the emotional tendency model is adjusted until the difference in pleasure between the response emotional data and the overall emotional data is less than 0.

5.

10. The digital human interaction system based on artificial intelligence optimization according to claim 9, characterized in that: The overall emotional data is emotional data of the overall image of the target digital person evaluated by a person's senses, and the overall emotional data is compared with the response data in terms of emotional pleasure; If the pleasure of the overall emotional data is greater than the pleasure of the response data by more than 0.5, then the emotional tendency model responds to the emotional data with a lower pleasure; If the pleasure of the overall emotional data is less than the pleasure of the response data by less than 0.5, then the emotional tendency model is increased to respond to the pleasure of the emotional data.

Images