Head portrait generation method and device, electronic equipment, medium and product

Abstract

The invention provides a head portrait generation method and device, electronic equipment, a medium and a product. According to one example of the invention, the method comprises the following steps: identifying a head portrait description audio recorded by a user to obtain text data corresponding to the head portrait description audio; inputting the text data into the large language model to obtain head portrait description information output by the large language model; and inputting the head portrait description information into a head portrait generation model to obtain a head portrait generated by the head portrait generation model.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a head portrait generation method, device, electronic device, medium and product. Background Art

[0002] With the rise of the Internet, various terminal devices and the applications in them cover multiple fields such as social networking, shopping, travel, education, and entertainment, greatly satisfying the diverse needs of users.

[0003] When using these terminal devices and applications, users often need to create personal accounts to enjoy personalized services or participate in community interactions. Setting up avatars is an important part of personal accounts. It not only represents the user's personal image, but also is an important way for users to show themselves and express their personality in the virtual world.

[0004] However, despite the widespread need to set up avatars, the implementation of avatar generation functions is relatively simple and lacks innovation. The current avatar generation methods are often limited to selecting pictures from albums, taking photos and uploading them, or using the default avatars built into the application. This single generation method does not provide rich and personalized user experience. Summary of the invention

[0005] In order to overcome the problems existing in the related art, the present application provides a head portrait generation method, device, electronic device, medium and product.

[0006] According to a first aspect of any embodiment of the present application, a method for generating an avatar is provided, the method comprising:

[0007] Recognize the avatar description audio recorded by the user to obtain text data corresponding to the avatar description audio;

[0008] Inputting the text data into a large language model to obtain avatar description information output by the large language model;

[0009] The avatar description information is input into an avatar generation model to obtain an avatar generated by the avatar generation model.

[0010] According to a second aspect of any embodiment of the present application, there is provided an avatar generation device, the device comprising:

[0011] An audio recognition module is used to recognize the avatar description audio recorded by the user to obtain text data corresponding to the avatar description audio;

[0012] An information acquisition module, used to input the text data into a large language model to obtain avatar description information output by the large language model;

[0013] The avatar acquisition module is used to input the avatar description information into the avatar generation model to obtain the avatar generated by the avatar generation model.

[0014] According to a third aspect of any embodiment of the present application, there is provided an electronic device, including:

[0015] processor;

[0016] a memory for storing processor-executable instructions;

[0017] The processor implements the method described in any embodiment of the present application by running the executable instructions.

[0018] According to a fourth aspect of any embodiment of the present application, a computer-readable storage medium is provided, on which computer instructions are stored. When the instructions are executed by a processor, the method described in any embodiment of the present application is implemented.

[0019] According to a fifth aspect of any embodiment of the present application, a computer program product is provided, on which a computer program / instruction is stored. When the computer program / instruction is executed by a processor, the method described in any embodiment of the present application is implemented.

[0020] The technical solution provided by this application may have the following beneficial effects:

[0021] According to the above embodiments, by recognizing the avatar description audio recorded by the user, text data corresponding to the avatar description audio is obtained, the text data is input into the large language model, the avatar description information output by the large language model is obtained, and the avatar description information is input into the avatar generation model to obtain the avatar generated by the avatar generation model, allowing the user to express the avatar description information in a more free and creative way, so that the generated avatar is close to the user's personal preferences and customization requirements, thereby obtaining a personalized, customized and interesting avatar.

[0022] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The accompanying drawings, which are incorporated in the specification and constitute a part of this application, illustrate embodiments consistent with the application and, together with the description, serve to explain the principles of the application.

[0024] Figure 1 is a flow chart of a method for generating an avatar according to an exemplary embodiment of the present application;

[0025] Figure 2 is a flowchart of another avatar generation method according to an exemplary embodiment of the present application;

[0026] Figure 3 is a schematic diagram of the structure of a terminal device according to an exemplary embodiment of the present application;

[0027] Figure 4 is a structural schematic diagram of an electronic device according to an exemplary embodiment of the present application;

[0028] Figure 5 It is a block diagram of an avatar generating device according to an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0029] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0030] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0031] It should be understood that although the terms first, second, third, etc. may be used in the present application to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0032] The current avatar generation method is limited to selecting pictures from the album, taking photos and uploading them, or using the default avatar built into the application, etc. This single generation method cannot meet the personalized needs of users.

[0033] In order to solve the above problems, the present application proposes a head portrait generation method. To further illustrate the present application, the following embodiments are provided:

[0034] See also Figure 1 , Figure 1This is a flowchart of a method for generating an avatar according to an exemplary embodiment of the present application. The avatar generating method may be executed by a terminal device or a server. The terminal device may be a computer, a tablet computer, a smart phone, a personal digital assistant, a learning machine or other electronic device, and the server may be a cloud server or the like.

[0035] Taking the execution subject as a terminal device as an example, the avatar generation method may include the following steps:

[0036] Step 102: Recognize the avatar description audio recorded by the user to obtain text data corresponding to the avatar description audio.

[0037] In this step, the user enters the avatar generation entrance of the personal account in the terminal device, for example, enters the "My" page in the application, clicks on the avatar or chooses to change the avatar.

[0038] The user can record the avatar description audio in real time or upload the pre-recorded avatar description audio. The terminal device receives the avatar description audio recorded by the user.

[0039] The terminal device can use speech recognition technologies such as automatic speech recognition (ASR) and deep learning technology to recognize the avatar description audio and obtain text data corresponding to the avatar description audio.

[0040] The avatar description audio is audio content that describes the avatar desired by the user. The avatar description audio may include a detailed description of the avatar's features, style, expression, hairstyle, clothing, etc., and may also include an imaginary description of a specific scene or atmosphere.

[0041] The text data is information extracted from the avatar description audio and converted into text form.

[0042] When the user is recording the avatar description audio in real time, the terminal device responds to detecting the user's recording start operation by recording the user's voice, and responds to detecting that the recording end condition is met by stopping recording the audio.

[0043] Exemplarily, the recording start operation may be a user clicking a start button, and the recording end condition may be detecting the user clicking the start button again, or the recording end condition may be detecting that the recording duration reaches a preset duration, for example: 30s, 60s, etc.

[0044] The recording start operation may also be a long press operation of the user on the start button, and the recording end condition may be detecting an end operation of the long press operation by the user.

[0045] Step 104: Input the text data into the large language model to obtain the avatar description information output by the large language model.

[0046] In this step, the terminal device can input the text data into the large language model. Based on its huge corpus and complex neural network structure, the large language model can understand and parse the semantic information in the text, perform error correction and polishing on the text data, adjust the colloquial or non-standard expressions in the text data, and generate richer and more accurate avatar description information. The terminal device obtains the avatar description information output by the large language model.

[0047] Before inputting text data into the large language model, the terminal device may pre-process the text data, such as removing irrelevant characters in the text data, so as to improve the efficiency and accuracy of the large language model processing.

[0048] The avatar description information is the description text of the avatar that the large language model expects to generate for the user.

[0049] A large language model is a type of deep learning model with a large number of parameters. It learns language patterns, grammar, and semantics by processing large amounts of text, thereby understanding and generating human language. It can be a chat generative pre-trained encoder-decoder model (Chat Generative Pre-Trained Transformer, ChatGPT), a pre-trained language model (Bidirectional Encoder Representations from Transformers, BERT), etc.

[0050] Step 106: Input the avatar description information into the avatar generation model to obtain the avatar generated by the avatar generation model.

[0051] In this step, the terminal device can input the avatar description information into the avatar generation model as a prompt of the avatar generation model. The avatar generation model generates corresponding image features such as color, shape, texture, etc. based on the input avatar description information, converts the image features into specific image data, and generates an avatar with clear details and realistic effects.

[0052] The terminal device obtains the avatar generated by the avatar generation model and can use the avatar as the user's avatar in the personal account.

[0053] Among them, the avatar generation model is used to generate an avatar that conforms to the avatar description information, which can be an image generation and editing model (A Discriminative Framework for Image-to-Image Translation, DALL.E), a stable diffusion model (Stable Diffusion, SD), etc.

[0054] The avatar generation model can also simultaneously output multiple avatars that conform to the avatar description information. The terminal device displays the multiple avatars to the user. In response to the user's selection of an avatar from the multiple avatars, the avatar is used as the user's avatar in the personal account.

[0055] The avatar generation method of this embodiment recognizes the avatar description audio recorded by the user to obtain text data corresponding to the avatar description audio, inputs the text data into a large language model to obtain avatar description information output by the large language model, inputs the avatar description information into an avatar generation model to obtain an avatar generated by the avatar generation model, allowing the user to express the avatar description information in a more free and creative way, so that the generated avatar is close to the user's personal preferences and customization requirements, thereby obtaining a personalized, customized and interesting avatar.

[0056] In addition, by recognizing the avatar description audio to obtain text data, users do not need to make detailed text description customization requirements, thus reducing the user's interaction cost. At the same time, combined with the powerful speech processing capabilities of the large language model, it can process the colloquial or non-standard content in the avatar description audio expressed by the user, generate more accurate avatar description text, and optimize the input problem of the next avatar generation model, so that the generated avatar is more in line with user needs.

[0057] In the above-mentioned embodiment, it is introduced that a large language model is used to obtain standardized avatar description information corresponding to the avatar description audio to generate a personalized avatar. In the following embodiment, how to generate an avatar that better meets the personalized needs of the user will be described in detail, and it can be applied to any of the above embodiments.

[0058] In one embodiment, the terminal device can obtain the user's historical avatars or a customized picture uploaded by the user to obtain a reference picture. The avatar description information and the reference picture are input into the avatar generation model to obtain the avatar generated by the avatar generation model.

[0059] The reference picture is used to provide guidance and reference for the avatar generation model. The reference picture may include at least one of the following: a user's historical avatar, or a picture uploaded by the user.

[0060] The historical avatars are avatars used by the user, and may be a preset number of avatars that the user has used recently, or may be all avatars that all users have used. This embodiment of the present application does not impose any limitation on this.

[0061] As described above, by obtaining the user's reference picture, inputting the avatar description information and the user's historical avatars into the avatar generation model, and obtaining the avatar generated by the avatar generation model, a personalized avatar that fits the style of the historical avatar can be obtained; by inputting the avatar description information and the picture uploaded by the user into the avatar generation model, and obtaining the avatar generated by the avatar generation model, multimodal data can be used to generate the avatar, so that the avatar is more in line with the user's personalized needs.

[0062] In one embodiment, the terminal device can judge the artistic style desired by the user based on the style description information in the text data, and determine the role played by the illustrator, cartoonist, etc. of the large language model.

[0063] The played role is used as guidance information of the large language model, and the large language model is guided to generate avatar description information with a specific artistic style as the played role. The terminal device obtains the avatar description information output by the large language model based on the guidance information.

[0064] The style description information is specific content that describes and specifies the artistic style of the avatar that the user expects, for example: the user expects the avatar to have an illustration style, with bright colors and imagination; the user expects the avatar to look like it is drawn by a cartoonist, with simple lines, etc.

[0065] The role played is the role played by the large language model when generating avatar description information, such as illustrator, cartoonist, etc.

[0066] The guidance information is a prompt used to guide the large language model to play a role. The guidance information may also include specific keywords or text templates and other information to assist the large language model in understanding the role to be played.

[0067] The terminal device may also display a plurality of preset role-playing characters to the user, and the user may select a desired role-playing character, and the role-playing character selected by the user may be used as guidance information.

[0068] As described above, by determining the role played by the large language model based on the style description information in the text data, and using the role played as guidance information for the large language model, the avatar description information output by the large language model based on the guidance information is obtained. The large language model can be designated as an illustrator, cartoonist, or other role to express the avatar description information, making the avatar description information more accurate and professional, and ensuring that the generated avatar conforms to the style description information.

[0069] In one embodiment, the terminal device can analyze the emotional features in the avatar description audio through audio analysis technologies such as emotion recognition algorithms and speech recognition technology, and input the text data and emotional features into the large language model.

[0070] The large language model generates avatar description information with corresponding emotional colors based on the text description and emotional characteristics in the input data. The terminal device obtains the avatar description information output by the large language model, which may include a detailed description of the avatar's expression, atmosphere, color, etc. to reflect the emotional characteristics expected by the user.

[0071] Among them, emotional features are emotional states extracted from the avatar description audio, such as: happiness, sadness, anger, calmness, etc.

[0072] As described above, by analyzing the emotional features in the avatar description audio, inputting the text data and emotional features into the large language model, and obtaining the avatar description information output by the large language model, the large language model can be made to output avatar description information with emotional colors, so that the generated avatar also has emotional colors.

[0073] In one embodiment, the terminal device can obtain the user's data profile and merge the user's data profile with text data.

[0074] The terminal device matches and integrates the key features and preferences in the data portrait with the specific avatar requirements in the text data to obtain fused data, and inputs the fused data into the large language model.

[0075] The big language model conducts in-depth understanding and analysis of the fused data, taking into account the personal characteristics and preferences reflected in the data portraits, as well as the specific avatar requirements in the text data, so that the generated avatar description information is more personalized and in line with user expectations. The terminal device obtains the avatar description information output by the big language model.

[0076] Among them, the fusion data is an input data set that integrates the user's personal characteristics and preferences, as well as the specific avatar requirements in the text data.

[0077] Data portrait is a comprehensive description of the user's personal characteristics and preferences. Data portrait can include information such as the user's age, gender, region, occupation, interests and hobbies.

[0078] As described above, by fusing the user's data portrait and text data to obtain fused data, the fused data is input into a large language model to obtain avatar description information output by the large language model, so that the avatar description information is more closely matched with the user's personal characteristics and preferences, ensuring that the generated avatar is more in line with the user's self-perception and expectations.

[0079] In one embodiment, the terminal device may receive the user's feedback audio on the last generated avatar, and use speech recognition technology to recognize the received user's feedback audio to obtain feedback text corresponding to the feedback audio.

[0080] The terminal device inputs the feedback text and the text data used to generate the avatar last time into the large language model. The large language model refers to the user's comments and suggestions in the feedback text, combines the specific requirements in the text data, and generates new avatar description information that better meets the user's expectations. The terminal device obtains the new avatar description information output by the large language model.

[0081] The terminal device inputs the new avatar description information into the avatar generation model, instructs the avatar generation model to generate a new avatar, and obtains a new avatar generated by the avatar generation model that better meets the user's expectations.

[0082] The feedback audio is used to reflect the user's evaluation information on the last generated avatar, and the feedback audio may include: the user's evaluation, suggestions or modification opinions on the avatar.

[0083] Feedback text is information extracted from feedback audio and converted into text form, which is used to describe the user's specific requirements or improvement suggestions for the new avatar.

[0084] As described above, by recognizing the feedback audio received from the user, the feedback text is obtained, the feedback text and text data are input into the large language model, the new avatar description information output by the large language model is obtained, the new avatar description information is input into the avatar generation model, and the new avatar generated by the avatar generation model is obtained, so as to realize multiple rounds of audio conversations with the user, utilize the user's feedback audio to continuously optimize and generate avatars that better meet the user's expectations, and improve the interactivity and satisfaction of the user with the terminal device.

[0085] To further introduce the avatar generation process, Figure 2 FIG. 4 shows a flow chart of another avatar generation method. Figure 2 As shown, the avatar generation method may include the following steps:

[0086] Step 202: Receive the avatar description audio recorded by the user.

[0087] In this step, the terminal device receives the avatar description audio recorded by the user.

[0088] See also Figure 3 , Figure 3 The schematic diagram of the structure of a terminal device is shown. Exemplarily, the terminal device 30 may include: a recording module 301, a speech-to-text module 302, a pre-processing module 303 and a model calling module 304.

[0089] The recording and receiving module 301 starts recording in response to the user clicking the start button. When the recording time of the avatar description audio does not reach 30s, the user clicking the start button is detected again, the recording is ended, and the avatar description audio is obtained.

[0090] If the recording time of the avatar description audio reaches 30 seconds and the user's click operation on the start button is not detected again, the recording is automatically ended to obtain the avatar description audio.

[0091] Step 204: Recognize the avatar description audio to obtain text data corresponding to the avatar description audio.

[0092] In this step, taking ASR as an example, the speech-to-text technology, the speech-to-text module 302 uses the ASR interface to recognize the avatar description audio to obtain text data corresponding to the avatar description audio.

[0093] Step 206: Use the role-playing as guidance information for the large language model.

[0094] In this step, the preprocessing module 303 determines the role played by the large language model according to the style description information in the text data, and writes the guidance information of the large language model based on the determined role played.

[0095] The preprocessing module 303 can also analyze the emotional features in the avatar description audio, and fuse the user's data portrait and text data to obtain fused data.

[0096] Step 208: input the text data into the large language model to obtain the avatar description information output by the large language model according to the guidance information.

[0097] In this step, the model calling module 304 can input text data into the large language model, and can also input fusion data and emotional features into the large language model to obtain the avatar description information output by the large language model according to the guidance information.

[0098] Step 210: Determine whether to use the reference picture to generate the image.

[0099] In this step, the terminal device can ask the user whether to use the reference picture to generate the image, and the user can choose whether to generate the image based on the avatar description audio only, or to generate the image based on the avatar description audio and the reference picture.

[0100] The model calling module 304 determines whether to use the reference picture to generate the image according to the received query result.

[0101] If the reference picture is not used to generate the image, then continue to step 212;

[0102] If the image is generated using a reference picture, step 214 is executed.

[0103] Step 212: Input the avatar description information into the avatar generation model.

[0104] In this step, the model calling module 304 inputs the avatar description information into the avatar generation model, for example, inputs the avatar description information into the prompt of the DALL.E model.

[0105] Step 214: Input the avatar description information and the reference image into the avatar generation model.

[0106] In this step, the model calling module 304 inputs the avatar description information and the reference image into the avatar generation model, for example, inputs the avatar description information and the reference image into the prompt of the SD model.

[0107] Step 216: Obtain the avatar generated by the avatar generation model.

[0108] In this step, the model calling module 304 obtains the avatar generation model to generate an avatar that meets the user's personalized needs.

[0109] Step 218: Recognize the feedback audio received from the user to obtain feedback text.

[0110] In this step, the audio recording module 301 can receive the user's feedback audio on the generated avatar, and the speech-to-text module 302 recognizes the feedback audio received from the user to obtain a feedback text.

[0111] Step 220: Input the feedback text and text data into the large language model to obtain new avatar description information output by the large language model.

[0112] In this step, the model calling module 304 inputs the feedback text and text data into the large language model to obtain new avatar description information output by the large language model that better meets the user's expectations.

[0113] Step 222: Input the new avatar description information into the avatar generation model to obtain a new avatar generated by the avatar generation model.

[0114] In this step, the model calling module 304 inputs the new avatar description information into the avatar generation model to obtain a new avatar generated by the avatar generation model. If the user is still not satisfied with the new avatar, he can give feedback audio again, and the terminal device repeats steps 218 to 222, and uses the user's feedback audio to continuously optimize and generate an avatar that better meets the user's expectations.

[0115] Figure 41 is a schematic diagram of the structure of an electronic device according to an exemplary embodiment of the present application. The electronic device may be, for example, a mobile phone, a computer, a digital broadcast terminal, a message transceiver, a game console, a tablet device, a personal digital assistant, a server, a smart home appliance, a car computer, etc. Figure 4 At the hardware level, the electronic device includes a processor 402, an internal bus 404, a network interface 406, a memory 408, and a non-volatile memory 410, and may also include hardware required for other services. The processor 402 reads the corresponding computer program from the non-volatile memory 410 into the memory 408 and then runs it, forming an avatar generation device at the logical level. Of course, in addition to the software implementation, this application does not exclude other implementation methods, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0116] Figure 5 is a block diagram of a head portrait generating device according to an exemplary embodiment of the present application. Figure 5 The device may include: an audio recognition module 502, an information acquisition module 504 and an avatar acquisition module 506, wherein:

[0117] The audio recognition module 502 is used to recognize the avatar description audio recorded by the user to obtain text data corresponding to the avatar description audio;

[0118] The information acquisition module 504 is used to input the text data into the large language model to obtain the avatar description information output by the large language model;

[0119] The avatar acquisition module 506 is used to input the avatar description information into the avatar generation model to obtain the avatar generated by the avatar generation model.

[0120] In one example, the avatar acquisition module 506, when used to input the avatar description information into the avatar generation model to obtain the avatar generated by the avatar generation model, includes: obtaining a reference picture of the user; wherein the reference picture includes at least one of the following: a historical avatar of the user, a picture uploaded by the user; inputting the avatar description information and the reference picture into the avatar generation model to obtain the avatar generated by the avatar generation model.

[0121] In one example, the information acquisition module 504, when used to input the text data into a large language model to obtain the avatar description information output by the large language model, includes: determining the role played by the large language model based on the style description information in the text data; using the role played as guidance information for the large language model, and obtaining the avatar description information output by the large language model based on the guidance information.

[0122] In one example, the information acquisition module 504, when used to input the text data into a large language model to obtain the avatar description information output by the large language model, includes: analyzing the emotional features in the avatar description audio; inputting the text data and the emotional features into the large language model to obtain the avatar description information output by the large language model.

[0123] In one example, the information acquisition module 504, when used to input the text data into a large language model to obtain the avatar description information output by the large language model, includes: fusing the user's data portrait and the text data to obtain fused data; and inputting the fused data into the large language model to obtain the avatar description information output by the large language model.

[0124] In one example, the audio recognition module 502 is also used to recognize the feedback audio received from the user to obtain feedback text; wherein the feedback audio is used to reflect the user's evaluation information on the avatar; the information acquisition module 504 is also used to input the feedback text and the text data into the large language model to obtain new avatar description information output by the large language model; the avatar acquisition module 506 is also used to input the new avatar description information into the avatar generation model to obtain a new avatar generated by the avatar generation model.

[0125] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0126] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present application scheme. A person of ordinary skill in the art can understand and implement it without paying creative labor.

[0127] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory including instructions. The instructions can be executed by a processor of the avatar generation device to implement any method as described in the above embodiments.

[0128] The non-temporary computer-readable storage medium may be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc., and the present application does not limit this.

[0129] In an exemplary embodiment, a computer program product including a computer program / instruction is also provided. The computer program / instruction can be executed by a processor of an avatar generating device to implement any of the methods described in the above embodiments.

[0130] The above describes specific embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0131] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the inventions claimed herein. The present application is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

[0132] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for generating an avatar, It is characterized in that The method comprises: Recognize the avatar description audio recorded by the user to obtain text data corresponding to the avatar description audio; Inputting the text data into a large language model to obtain avatar description information output by the large language model; The avatar description information is input into an avatar generation model to obtain an avatar generated by the avatar generation model.

2. The method according to claim 1, It is characterized in that The step of inputting the avatar description information into an avatar generation model to obtain an avatar generated by the avatar generation model includes: Obtain a reference picture of the user; wherein the reference picture includes at least one of the following: a historical profile picture of the user, a picture uploaded by the user; The avatar description information and the reference image are input into the avatar generation model to obtain the avatar generated by the avatar generation model.

3. The method according to claim 1, It is characterized in that The step of inputting the text data into a large language model to obtain avatar description information output by the large language model includes: Determining the role played by the large language model according to the style description information in the text data; The role-playing is used as guidance information of the large language model to obtain avatar description information output by the large language model according to the guidance information.

4. The method according to claim 1, It is characterized in that The step of inputting the text data into a large language model to obtain avatar description information output by the large language model includes: Analyzing emotional features in the avatar description audio; The text data and the emotion feature are input into the large language model to obtain the avatar description information output by the large language model.

5. The method according to claim 1, It is characterized in that The step of inputting the text data into a large language model to obtain avatar description information output by the large language model includes: Fusing the user's data profile with the text data to obtain fused data; The fused data is input into the large language model to obtain the avatar description information output by the large language model.

6. The method according to claim 1, It is characterized in that The method further comprises: Recognize the feedback audio received from the user to obtain feedback text; wherein the feedback audio is used to reflect the user's evaluation information on the avatar; Inputting the feedback text and the text data into the large language model to obtain new avatar description information output by the large language model; The new avatar description information is input into the avatar generation model to obtain a new avatar generated by the avatar generation model.

7. A head portrait generating device, It is characterized in that The device comprises: An audio recognition module is used to recognize the avatar description audio recorded by the user to obtain text data corresponding to the avatar description audio; An information acquisition module, used to input the text data into a large language model to obtain avatar description information output by the large language model; The avatar acquisition module is used to input the avatar description information into the avatar generation model to obtain the avatar generated by the avatar generation model.

8. An electronic device, It is characterized in that include: processor; a memory for storing processor-executable instructions; The processor implements the method according to any one of claims 1 to 6 by running the executable instructions.

9. A computer-readable storage medium having computer instructions stored thereon, It is characterized in that When the instruction is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer program product having a computer program / instructions stored thereon, It is characterized in that When the computer program / instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images