Enterprise training dialogue interaction method and system based on virtual digital human

Abstract

The invention provides an enterprise training dialogue interaction method and system based on a virtual digital human, and relates to the technical field of artificial intelligence, and the method comprises the steps: obtaining a virtual character description prompt word, and generating a virtual character image according to the description prompt word; based on the virtual character image, utilizing an action migration algorithm to migrate the action of the real character to the virtual character image to form an action set, and constructing a virtual digital person; user inquiry audio information is input into the virtual digital person, the audio information is converted into text information through a voice recognition algorithm, the text information is analyzed through a large model, answers are output, and the virtual digital person synthesizes the answers of the text types output by the large model into voice with set timbres through a voice synthesis algorithm. A smoothing algorithm is introduced, the synthesized voice is used for driving the lip shape of the virtual digital person to change in real time, meanwhile, a corresponding action video is called from the action set of the virtual digital person according to a keyword triggering mode, the virtual digital person is made to complete actions, and the voice is played.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to a method and system for enterprise training dialogue interaction based on virtual digital humans. Background Art

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] With the rapid development of artificial intelligence technology, AI algorithms have made excellent progress in image generation, speech synthesis, speech recognition and other fields. The progress of artificial intelligence technology makes it possible to apply virtual digital human systems in the fields of e-commerce live broadcast, cultural tourism and vocational training.

[0004] The virtual digital human system consists of a virtual digital human and related supporting hardware. To make a virtual digital human, you first need to create a virtual character, and then use algorithms such as lip drive, motion drive, speech synthesis, and speech recognition to drive the virtual character, so that the virtual character has the ability to communicate and interact with users. The supporting hardware related to the virtual digital human system includes voice input devices, voice output devices, high-performance computing servers, and displays.

[0005] Currently, virtual digital humans have been used in the field of corporate training. Through interactive dialogues between virtual characters and corporate personnel, the needs of multiple corporate employees to conduct real-time training with digital humans online at the same time are met.

[0006] However, the application of traditional virtual digital humans in the field of corporate training still has the following shortcomings:

[0007] 1. The virtual digital human has poor lip shape driving effect. The synthesized lips usually have a different resolution from the original virtual character's face, and there is a clear sense of separation between the lips and the face, which affects the experience of corporate employees interacting with the digital human;

[0008] 2. The virtual digital human has a low level of intelligence and can only answer simple questions set in advance. It cannot accurately answer the company's training content and therefore cannot meet the company's training needs;

[0009] 3. The real-time performance of the virtual digital human system is poor. It can only communicate and interact with a single user at the same time. It cannot allow multiple people to interact online at the same time, and cannot meet the task of large-scale personnel training in enterprises. Summary of the invention

[0010] In order to solve the above problems, the present invention proposes a method and system for enterprise training dialogue interaction based on virtual digital humans, optimizes the lip shape driving algorithm of virtual digital humans, combines virtual digital humans with large models, proposes a large model fine-tuning data processing method for enterprise training, and adds WebRTC technology to the virtual digital human system to meet the needs of multiple enterprise employees to conduct real-time training with digital humans online at the same time.

[0011] According to some embodiments, the present disclosure adopts the following technical solutions:

[0012] The enterprise training dialogue interaction method based on virtual digital human includes:

[0013] Obtaining description prompt words for the virtual character, and generating an image of the virtual character according to the description prompt words;

[0014] Based on the virtual character image, the action transfer algorithm is used to transfer the real person's actions to the virtual character image to form an action set and build a virtual digital human;

[0015] The audio information of the user's inquiry is input into the virtual digital human, and the audio information is converted into text information using a speech recognition algorithm. The text information is analyzed using a large model and the answer is output. The virtual digital human uses a speech synthesis algorithm to synthesize the answer into a speech with a set timbre based on the text type output by the large model. A smoothing algorithm is introduced to use the synthesized speech to drive the changes in the virtual digital human's lip shape in real time. At the same time, the corresponding action video is called from the virtual digital human's action set according to the triggering keyword, so that the virtual digital human completes the action and plays the voice.

[0016] According to some embodiments, the present disclosure adopts the following technical solutions:

[0017] The enterprise training dialogue interaction system based on virtual digital human includes:

[0018] A virtual human construction module is used to obtain virtual human description prompt words and generate a virtual human image according to the description prompt words; based on the virtual human image, using the action transfer algorithm, the actions of the real person are transferred to the virtual human image to form an action set, and a virtual digital human is constructed;

[0019] The interactive module is used to input the audio information of the user's inquiry into the virtual digital human, use the speech recognition algorithm to convert the audio information into text information, use the large model to analyze the text information and output the answer, and the virtual digital human uses the speech synthesis algorithm to synthesize the answer of the text type output by the large model into a speech with a set timbre. A smoothing algorithm is introduced to use the synthesized speech to drive the changes in the lip shape of the virtual digital human in real time. At the same time, the corresponding action video is called from the virtual digital human action set according to the triggering keyword, so that the virtual digital human completes the action and plays the voice.

[0020] According to some embodiments, the present disclosure adopts the following technical solutions:

[0021] A computer program product includes a computer program, and when the computer program is executed by a processor, the enterprise training dialogue interaction method based on virtual digital human is implemented.

[0022] According to some embodiments, the present disclosure adopts the following technical solutions:

[0023] A non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the enterprise training dialogue interaction method based on virtual digital human is implemented.

[0024] According to some embodiments, the present disclosure adopts the following technical solutions:

[0025] An electronic device comprises: a processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the enterprise training dialogue interaction method based on virtual digital human.

[0026] Compared with the prior art, the present invention has the following beneficial effects:

[0027] The disclosed enterprise training dialogue interaction method based on virtual digital human generates a virtual character image; based on the virtual character image, the action transfer algorithm is used to transfer the actions of the real person to the virtual character image to form an action set, and a virtual digital human is constructed. The virtual digital human can be designed into various images according to the training content, such as experts, mentors or colleagues, and communicate with trainees through natural language processing technology. This interactive method is more attractive than traditional reading materials or watching videos, and can increase the participation and interest of trainees.

[0028] The disclosed enterprise training dialogue interaction method based on virtual digital humans utilizes a speech recognition algorithm to convert audio information into text information, utilizes a large model to analyze the text information and output answers, and the virtual digital humans provide answers of the text type based on the large model output. The virtual digital humans based on the large model can provide customized teaching plans based on the learning progress, preferences and abilities of each trainee, thereby helping trainees to master knowledge and skills more effectively and reduce training costs. In the long run, using a virtual digital human system for training can reduce the demand for physical locations and human resources, thereby reducing training costs.

[0029] The disclosed enterprise training dialogue interaction method based on virtual digital human uses a speech synthesis algorithm to synthesize a speech with a set tone, introduces a smoothing algorithm to use the synthesized speech to drive the change of the virtual digital human's lip shape in real time, first generates a static virtual digital human image according to the prompt word, and then uses the action migration algorithm to drive the static virtual digital human into a dynamic video, each video corresponding to an action. Then, according to the triggering keyword, the corresponding action video is called from the virtual digital human action set; when the digital human is answering a question, once a keyword is triggered, such as "introduction", "hello", etc., the system will automatically call the driven action video, so that the digital human completes the corresponding action, improving the interaction efficiency.

[0030] The disclosed enterprise training dialogue interaction method based on virtual digital humans can provide guidance and support to trainees 24 hours a day. Using the disclosed virtual digital human system for training can ensure that all employees receive the same quality of information transfer and skills training, which helps to maintain the consistency and efficiency of internal operations of the enterprise and standardize training content. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The accompanying drawings constituting a part of the present disclosure are used to provide a further understanding of the present disclosure. The illustrative embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation on the present disclosure.

[0032] Figure 1 It is a method flow architecture diagram of an embodiment of the present disclosure;

[0033] Figure 2 A schematic diagram of a virtual digital human according to an embodiment of the present disclosure;

[0034] Figure 3 A schematic diagram of a question-answer pair according to an embodiment of the present disclosure;

[0035] Figure 4 A schematic diagram of a multi-round question-answer pair according to an embodiment of the present disclosure;

[0036] Figure 5 The hardware architecture diagram of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0037] The present disclosure is further described below in conjunction with the accompanying drawings and embodiments.

[0038] It should be noted that the following detailed descriptions are all illustrative and are intended to provide further explanation of the present disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present disclosure belongs.

[0039] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present disclosure. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0040] Example 1

[0041] In one embodiment of the present disclosure, a method for enterprise training dialogue interaction based on a virtual digital human is provided, comprising:

[0042] Obtaining description prompt words for the virtual character, and generating an image of the virtual character according to the description prompt words;

[0043] Based on the virtual character image, the action transfer algorithm is used to transfer the real person's actions to the virtual character image to form an action set and build a virtual digital human;

[0044] The audio information of the user's inquiry is input into the virtual digital human, and the audio information is converted into text information using a speech recognition algorithm. The text information is analyzed using a large model and the answer is output. The virtual digital human uses a speech synthesis algorithm to synthesize the answer into a speech with a set timbre based on the text type output by the large model. A smoothing algorithm is introduced to use the synthesized speech to drive the changes in the virtual digital human's lip shape in real time. At the same time, the corresponding action video is called from the virtual digital human's action set according to the triggering keyword, so that the virtual digital human completes the action and plays the voice.

[0045] As an embodiment, the specific implementation process of a virtual digital human-based enterprise training dialogue interaction method disclosed in the present invention is as follows:

[0046] Step 1: Obtain description prompt words for the virtual character, and generate a virtual character image according to the description prompt words;

[0047] Specifically, a virtual character image is generated through the Stable Diffusion algorithm. The Stable Diffusion algorithm inputs text information, which (referred to as prompt words) describes the characteristics of the virtual character, and outputs a virtual character image that is similar to the description of the text information. The final virtual character image is obtained by continuously adjusting the prompt words. For example, if the prompt word "help me generate an image of a lady" is input, Stable Diffusion will generate an image of a girl. If you feel that the generated image does not meet your requirements, you can adjust the prompt words such as "help me generate a lady who is wearing a white top and blue shorts", etc. After repeated adjustments of the prompt words and continuous trial and error, a more satisfactory image is finally determined.

[0048] Step 2: Based on the virtual character image, use the action transfer algorithm to transfer the real person's actions to the virtual character image to form an action set and build a virtual digital human;

[0049] Specifically, after obtaining the image of the virtual character, the artificial intelligence action transfer algorithm is used to transfer the actions of the real person to the virtual character, so that the virtual character completes the corresponding actions, thereby obtaining videos of different actions to form an action set.

[0050] The method includes: firstly recording a video of a single action of a real person, and then inputting a virtual character image generated by Stable Diffusion and the real person action video into an action migration algorithm to obtain an action video of a virtual character.

[0051] By continuously changing the action videos of real people, multiple action videos of the same virtual character can be obtained, and these videos can be stored together as an action set for easy subsequent use.

[0052] Step 3: Input the audio information of the user's inquiry into the virtual digital human, use the speech recognition algorithm to convert the audio information into text information, use the big model to analyze the text information and output the answer, and the virtual digital human uses the speech synthesis algorithm to synthesize the answer into a speech with a set tone based on the text type output by the big model. Introduce a smoothing algorithm to use the synthesized speech to drive the changes in the virtual digital human's lip shape in real time. At the same time, call the corresponding action video from the virtual digital human action set according to the trigger keyword, so that the virtual digital human completes the action and plays the voice.

[0053] Specifically, the input device is used to collect the user's inquiry audio information, and the virtual digital human uses the existing speech recognition algorithm to convert the collected audio information into text information.

[0054] Furthermore, the text information is input into the big model, and the big model returns the answer to the virtual digital human in the form of text information.

[0055] The training and fine-tuning process of the large model is as follows:

[0056] 1. Collect documents related to corporate training. Training documents include various formats, such as ppt, pdf, word, excel, etc. Different types of files are classified for subsequent unified processing.

[0057] 2. Clean the file data. Mainly including: removing invalid information, removing duplicate information, removing erroneous information, processing missing values, extracting text from images, privacy protection, etc.

[0058] 3. Organize the cleaned data into the form of "question-answer" (Q&A pairs). In order to make the fine-tuning effect of the large model better, a large Q&A pair is split into multiple medium Q&A pairs, and the medium Q&A pairs can be further split into multiple small Q&A pairs.

[0059] 4. In order to improve the large model's ability to understand the content of the training file, multiple questioning methods are used for the same question, and the answers corresponding to different questioning methods are the same. This method can convert the same question and answer pair into multiple question and answer pairs, where the questioning methods of multiple question and answer pairs are different but the answers are the same.

[0060] 5. A single training sample in the training set consists of 3-6 question-answer pairs to ensure that the fine-tuned large model has the ability to conduct multiple rounds of dialogue. There are three ways to organize multiple question-answer pairs in a single sample:

[0061] ① Randomly select 3-6 question-answer pairs of different granularities from the same question to form a sample.

[0062] ②Mix the question-answer pairs of similar questions and randomly select 3-6 question-answer pairs to form a sample.

[0063] ③Mix the question-answer pairs of different questions, and randomly select 3-6 question-answer pairs to form a sample.

[0064] Preferably, in order to ensure that the fine-tuned large model does not lose its general capabilities, a portion of the general data set is selected and added to the training set. The general data set includes multiple rounds of daily conversations, law, medicine and other field knowledge.

[0065] Furthermore, the virtual digital human uses the existing speech synthesis algorithm to synthesize the text-type answer output by the large model into a speech with a set timbre.

[0066] Furthermore, the synthesized voice is used to drive the changes in the lip shape of the virtual digital human, and an optimized lip shape driving algorithm is used for real-time driving, and the driving results are processed in high definition, specifically:

[0067] The lip shape driving algorithm is used to drive the virtual digital human, so that the lip shape of the digital human changes accordingly during the speech process. The lip shape driving algorithm is optimized on the basis of wav2lip, and a smoothing algorithm is added to make the junction between the synthesized lip shape and the face smoother, improving the visual effect of the virtual digital human. The smoothing algorithm performs weighted fusion on the area to be smoothed to achieve a gradual in and out effect. The smoothing algorithm formula is:

[0068]

[0069] Among them, f(x,y) is the smoothed image, f1(x,y) is the lip shape image, f2(x,y) is the face image, w1 and w2 are the weights of the smoothing algorithm. The calculation formulas of w1 and w2 are as follows:

[0070]

[0071] Among them, x i is the current pixel position, x r is the right boundary of the smoothing region, x l is the left boundary of the smoothing region.

[0072] Furthermore, according to the answer fed back by the big model, the corresponding action video is called from the action set of the virtual digital human in the way of triggering keywords, so that the virtual digital human completes the action, and the synthesized voice is played out using the output device. 1) For example, the answer fed back by the big model is "Hello everyone, I am a virtual digital human." Then "Hello everyone" is a keyword, and the system will call the action video of "greeting" from the action set of the digital human, so that the digital human completes the action of greeting.

[0073] 2) For example, if the big model gives the answer "Today I will introduce a product to you", then "introduction" is a keyword, and the system will call the action video of "introduction" from the digital human action set to enable the digital human to complete the action of introduction.

[0074] As an embodiment, the characteristics of a virtual digital human multi-person online interaction disclosed in the present invention are:

[0075] 1. Support multiple users to connect using different platforms. Each connected user corresponds to a different digital person and does not interfere with each other.

[0076] 2. Each user will be assigned a unique room_id when connecting. This identifier is not only used to distinguish different user sessions, but can also be bound to the user's specified avatar, allowing users to switch between different avatars.

[0077] 3. The virtual digital human system uses a speech synthesis algorithm to synthesize speech, generates a virtual image and a variety of driving algorithms to synthesize video, and synchronizes audio and video based on timestamps.

[0078] 4. The results generated by the virtual digital human system are efficiently and stably transmitted to each user connected to the corresponding room_id using WebRTC technology. This not only ensures the security and efficiency of data transmission, but also greatly enhances the user's immersion and interactivity.

[0079] 5. Multiple users complete relevant content of corporate training through real-time communication with virtual digital people.

[0080] Example 2

[0081] In one embodiment of the present disclosure, a virtual digital human-based enterprise training dialogue interaction system is provided, comprising:

[0082] A virtual human construction module is used to obtain virtual human description prompt words and generate a virtual human image according to the description prompt words; based on the virtual human image, using the action transfer algorithm, the actions of the real person are transferred to the virtual human image to form an action set, and a virtual digital human is constructed;

[0083] The interactive module is used to input the audio information of the user's inquiry into the virtual digital human, use the speech recognition algorithm to convert the audio information into text information, use the large model to analyze the text information and output the answer, and the virtual digital human uses the speech synthesis algorithm to synthesize the answer of the text type output by the large model into a speech with a set timbre. A smoothing algorithm is introduced to use the synthesized speech to drive the changes in the lip shape of the virtual digital human in real time. At the same time, the corresponding action video is called from the virtual digital human action set according to the triggering keyword, so that the virtual digital human completes the action and plays the voice.

[0084] As an embodiment, the hardware of the enterprise training dialogue interaction system based on virtual digital human includes a display, a computer server, an audio input device and an audio output device. The display is used to display the virtual digital human. The virtual digital human is in a simple standing posture when not communicating and interacting with the user. When the virtual digital human gives feedback to the user, the lip shape changes and the corresponding action is completed.

[0085] The computer server includes a virtual human construction module and an interaction module, and is a key device of the virtual digital human system. It is used for the calculation of multiple algorithms and specifically executes the method described in Example 1.

[0086] The audio input device is used to collect the user's voice and transmit it to the voice recognition algorithm of the computer server. When the user needs to ask a question to the digital human, press and hold the collection button to start recording the sound. When the question is finished, release the collection button and the recording is finished.

[0087] The audio output device is used to play the voice synthesized by the virtual digital human, which is the answer to the user's question.

[0088] Example 3

[0089] In one embodiment of the present disclosure, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the enterprise training dialogue interaction method based on virtual digital human is implemented.

[0090] Example 4

[0091] In one embodiment of the present disclosure, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the enterprise training dialogue interaction method based on virtual digital human is implemented.

[0092] Example 5

[0093] In one embodiment of the present disclosure, an electronic device is provided, including: a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the enterprise training dialogue interaction method based on virtual digital human.

[0094] The present disclosure is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present disclosure. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0095] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0096] Although the above describes the specific implementation methods of the present disclosure in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present disclosure. Technical personnel in the relevant field should understand that on the basis of the technical solution of the present disclosure, various modifications or variations that can be made by those skilled in the art without creative work are still within the scope of protection of the present disclosure.

Claims

1. An enterprise training dialogue interaction method based on virtual digital human, characterized in that: include: Obtaining description prompt words for the virtual character, and generating an image of the virtual character according to the description prompt words; Based on the virtual character image, the action transfer algorithm is used to transfer the real person's actions to the virtual character image to form an action set and build a virtual digital human; The audio information of the user's inquiry is input into the virtual digital human, and the audio information is converted into text information using a speech recognition algorithm. The text information is analyzed using a large model and the answer is output. The virtual digital human uses a speech synthesis algorithm to synthesize the answer into a speech with a set timbre based on the text type output by the large model. A smoothing algorithm is introduced to use the synthesized speech to drive the changes in the virtual digital human's lip shape in real time. At the same time, the corresponding action video is called from the virtual digital human's action set according to the triggering keyword, so that the virtual digital human completes the action and plays the voice.

2. The enterprise training dialogue interaction method based on virtual digital human as claimed in claim 1, characterized in that: Obtain the description words of the virtual character, input the text information into the Stable Diffusion algorithm, extract the characteristics of the virtual character described in the text information, output the virtual character image that is similar to the description of the text information, and obtain the final virtual character image by continuously adjusting the characteristics of the text information.

3. The enterprise training dialogue interaction method based on virtual digital human as claimed in claim 1, characterized in that: After obtaining the image of the virtual character, the artificial intelligence action transfer algorithm is used to transfer the actions of the real person to the virtual character, so that the virtual character can complete the corresponding actions, thereby obtaining videos of different actions, forming an action set, and constructing a virtual digital person.

4. The enterprise training dialogue interaction method based on virtual digital human as claimed in claim 1, characterized in that: The input device is used to collect audio information of user questions, and the virtual digital human converts the collected audio information into text information using a speech recognition algorithm, and inputs the text information into a large model. The large model is trained in the form of question-answer pairs, and large question-answer pairs are split into multiple medium question-answer pairs. The medium question-answer pairs are further split into multiple small question-answer pairs, and the same question-answer pair is converted into multiple question-answer pairs. The multiple question-answer pairs have different question-ask methods but the same answers, to ensure that the large model after training and fine-tuning has the ability to have multiple rounds of conversations.

5. The enterprise training dialogue interaction method based on virtual digital human as claimed in claim 1, characterized in that: The lip shape driving algorithm is used to drive the virtual digital human, so that the lip shape of the digital human changes accordingly during speaking. The lip shape driving algorithm is optimized based on wav2lip and a smoothing algorithm is added. The smoothing algorithm formula is: Among them, f(x, y) is the smoothed image, f1(x, y) is the lip image, f2(x, y) is the face image, and w1 and w2 are the weights of the smoothing algorithm.

6. The enterprise training dialogue interaction method based on virtual digital human as claimed in claim 5, characterized in that: The calculation formulas for w1 and w2 are as follows: Among them, x i is the current pixel position, x r is the right boundary of the smoothing region, x l is the left boundary of the smoothing region.

7. The enterprise training dialogue interaction system based on virtual digital human is characterized by: include: A virtual person construction module is used to obtain description prompt words of the virtual person and generate a virtual person image according to the description prompt words; Based on the virtual character image, the action transfer algorithm is used to transfer the real person's actions to the virtual character image to form an action set and build a virtual digital human; The interactive module is used to input the audio information of the user's inquiry into the virtual digital human, use the speech recognition algorithm to convert the audio information into text information, use the large model to analyze the text information and output the answer, and the virtual digital human uses the speech synthesis algorithm to synthesize the answer of the text type output by the large model into a speech with a set timbre. A smoothing algorithm is introduced to use the synthesized speech to drive the changes in the lip shape of the virtual digital human in real time. At the same time, the corresponding action video is called from the virtual digital human action set according to the triggering keyword, so that the virtual digital human completes the action and plays the voice.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the enterprise training dialogue interaction method based on virtual digital human described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by the processor, the enterprise training dialogue interaction method based on virtual digital human as described in any one of claims 1-6 is implemented.

10. An electronic device, characterized in that: include: A processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the enterprise training dialogue interaction method based on virtual digital human as described in any one of claims 1-6.

Images