Advertisement generation method, device and equipment and readable storage medium

Abstract

The invention discloses an advertisement generation method, device and equipment and a readable storage medium, and the method comprises the steps: obtaining a target large model agent and a voice signal inputted by a user, and the agent comprises an intention recognition level, a frame generation level, a decision level and an editing level; based on the intention recognition level, intention recognition is carried out on the voice signal, and an advertisement strategy of the user is determined; generating a narrative framework based on the advertisement strategy by using the framework generation hierarchy; by utilizing the decision-making hierarchy, based on the narrative framework, obtaining each matched constituent element, and constructing a plurality of sub-mirror scripts; and based on the editing hierarchy, synthesizing each sub-mirror script to generate an advertisement. Therefore, the dependence on manual design in the advertisement generation process is reduced, the labor cost and the time cost are reduced, and the resource utilization efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and more specifically, to an advertisement generation method, apparatus, device, and readable storage medium. Background Art

[0002] With the rapid development of digital marketing, advertising has become a crucial means of selling goods. Traditional social media ad creation relies heavily on manual labor, from creative conception and content creation to delivery strategy formulation. This is both labor-intensive and time-consuming, and fails to meet advertisers' demand for efficient delivery. To address this issue, some companies are leveraging AI algorithms to assist in ad creation. Using machine learning to analyze user profile data, they can automatically generate ad copy and recommend creatives, significantly improving production efficiency.

[0003] However, existing technologies still have significant limitations. Existing AI ad generation algorithms often use fixed templates, resulting in monotonous and lacking creativity in generated ad copy and creative content. For example, ads for similar products often share highly similar content, resulting in poor ad playback performance and failing to meet advertisers' needs. Summary of the Invention

[0004] In view of this, the present application provides an advertisement generation method, apparatus, device and readable storage medium to address the shortcomings of the prior art in that advertisement content is severely homogenized.

[0005] In order to achieve the above objectives, the following solutions are proposed:

[0006] A method for generating an advertisement, comprising:

[0007] Obtain the target large model agent and the user's input voice signal, wherein the agent includes an intent recognition layer, a framework generation layer, a decision layer, and an editing layer;

[0008] Based on the intention recognition level, performing intent recognition on the voice signal to determine the user's advertising strategy;

[0009] generating a narrative framework based on the advertising strategy using the framework generation hierarchy;

[0010] Utilizing the decision-making hierarchy and based on the narrative framework, obtaining matching constituent elements and constructing multiple storyboards;

[0011] Based on the editing level, each storyboard script is synthesized to generate an advertisement.

[0012] Optionally, obtaining the target large model agent includes:

[0013] Obtaining an initial agent model and different training ads, and performing strategy extraction on each training ad to obtain strategy data for each training ad, wherein the initial agent model includes an initial intent recognition layer, an initial framework generation layer, an initial decision layer, and an initial editing layer;

[0014] Extract key features from the strategy data of each training ad to obtain multiple strategy subsets;

[0015] The strategy data and each strategy subset corresponding to the same training advertisement are used as strategy samples;

[0016] Segment each training advertisement to obtain multiple advertisement segments;

[0017] Based on each advertising segment of each training advertisement, a training narrative framework of each training advertisement is constructed;

[0018] Extract the components from each advertising clip, determine the recall keywords corresponding to each component, and determine the storyboard training script corresponding to each advertising clip;

[0019] The initial agent model is trained according to the strategy sample, training narrative framework and storyboard training script corresponding to each training advertisement until the initial agent model meets the preset stopping condition; the final initial agent model is used as the target large model agent.

[0020] Optionally, also include:

[0021] Obtain the playback effects of different types of social ads on different advertising platforms;

[0022] Obtain user feedback on different types of social ads on different advertising platforms;

[0023] Based on the playback effects and user feedback of different types of social advertisements on the same advertising platform, the parameters of the agent are optimized.

[0024] Optionally, performing intent recognition on the voice signal based on the intent recognition level and determining the user's advertising strategy includes:

[0025] Based on the intent recognition level, convert the voice signal into text to obtain text information, and extract keywords from the text information to determine key fields containing advertising style and advertising content planning;

[0026] Combined with the preset advertisement structure formula library, an advertisement strategy matching the key fields is generated.

[0027] Optionally, utilizing the decision-making hierarchy and based on the narrative framework to obtain matching constituent elements and construct multiple storyboards includes:

[0028] By utilizing the decision-making hierarchy and combining it with the advertising strategy, the narrative framework is disassembled to obtain multiple storyboard overviews, and content decisions are made for each storyboard overview to determine the picture requirements, lens language, sound effect requirements and storyboard emotions, and generate a storyboard script.

[0029] Optionally, the content decision-making for each storyboard overview, determining the image requirements, shot language, sound effect requirements, and storyboard emotions, and generating a storyboard script may include:

[0030] The decision-making hierarchy is used to make content decisions for each storyboard overview, determine the picture requirements, lens language, sound effect requirements and storyboard emotions, and perform multi-dimensional recall of video materials based on the picture requirements and lens language corresponding to each storyboard overview, obtain storyboard videos that match the picture requirements and lens language, and select matching storyboard sound effects and storyboard timbre based on the sound effect requirements and storyboard emotions corresponding to each storyboard overview, and generate storyboard lines based on each storyboard overview and its corresponding picture requirements, lens language, and storyboard emotions; the storyboard videos, storyboard sound effects, storyboard timbre and storyboard lines of the same storyboard overview constitute the corresponding storyboard script.

[0031] Optionally, synthesizing the storyboards based on the editing level to generate an advertisement includes:

[0032] By utilizing the editing level, for each storyboard script, with the goal of maximizing the degree of match between the storyboard overview of the storyboard script and each constituent element, the screen content within the storyboard video of the storyboard script is adjusted, the storyboard timbre and storyboard sound effects of the storyboard script are attribute adjusted, the storyboard lines of the storyboard script are optimized, and the optimized storyboard lines are presented with the adjusted storyboard timbre to obtain storyboard voice, and the adjusted storyboard video, storyboard sound effects and storyboard voice are structured and synthesized to obtain an advertisement.

[0033] An advertisement generating device, comprising:

[0034] The acquisition module is used to obtain the target large model agent and the voice signal input by the user. The agent includes the intention recognition layer, the framework generation layer, the decision layer and the editing layer;

[0035] a determination module, configured to perform intent recognition on the voice signal based on the intent recognition level and determine the user's advertising strategy;

[0036] a generating module for generating a narrative framework based on the advertising strategy by using the framework to generate a hierarchy;

[0037] A construction module, configured to utilize the decision-making hierarchy and, based on the narrative framework, obtain matching constituent elements and construct a plurality of storyboards;

[0038] A synthesis module is used to synthesize various storyboards based on the editing level to generate advertisements.

[0039] An advertisement generating device comprising a memory and a processor;

[0040] The memory is used to store programs;

[0041] The processor is used to execute the program to implement each step of the above-mentioned advertisement generation method.

[0042] A readable storage medium stores a computer program, which, when executed by a processor, implements the various steps of the above-mentioned advertisement generation method.

[0043] It can be seen from the above technical solutions that the advertisement generation method provided by the present application can obtain the voice signal input by the user, use the agent's intention recognition level to perform intent recognition on the voice signal, and determine the user's advertising strategy; based on this, the present application can interact with the user through the user's voice input, provide the user with a convenient way to input advertising needs, and enhance the user's participation in advertising editing and the convenience of advertising generation operations; determine the advertising strategy based on the voice input of each user, so that the advertising strategy is personalized and differentiated, thereby avoiding the homogenization of advertising content while making the advertisement more in line with user needs, improving the relevance of the advertising slogan to the target audience, and improving the advertising playback effect; on this basis, using the framework generation level, based on the advertising strategy, a narrative framework is generated; using the decision-making level, based on the narrative framework, the matching components are obtained to construct multiple storyboards; based on the editing level, the storyboards are synthesized to generate advertisements. Based on this, the present application can generate a narrative framework in accordance with the advertising strategy in a targeted manner, and generate multiple storyboards based on the narrative framework. The narrative framework can ensure the coherence and logic between the various storyboards, decompose the narrative framework into multiple storyboards, enrich the creative expression and form of the advertisement, greatly reduce the probability of advertisement homogeneity, and convert the entire advertisement generation process into a determination process and synthesis process of multiple storyboards, thereby simplifying the difficulty of advertisement generation; recalling the corresponding constituent elements of the narrative framework and then forming a storyboard can make the advertisement content fit the needs of different advertising stages, enhance the overall coordination and expressiveness of the advertisement, and enrich the content contained in the advertisement, thereby avoiding advertisement homogeneity. Advertising homogeneity; It can be seen that the target large model agent of this application has a clear division of labor at each level, and the voice interaction method and each level work closely together to provide users with convenient, efficient and personalized advertising generation services, improve user participation and satisfaction in the advertising production process, and generate advertisements based on user needs to ensure that the generated advertisements are unique and avoid stereotyped advertising content; process-based operations ensure efficient advertising production, and at the same time, professional-level processing improves the logic, coherence and overall quality of advertising content; users input voice signals to obtain advertisements that meet their needs, reduce dependence on manual design in the advertising generation process, reduce labor costs and time costs, and improve resource utilization efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without any creative work.

[0045] Figure 1 A flow chart of an advertisement generation method disclosed in an embodiment of the present application;

[0046] Figure 2 This is a structural block diagram of an advertisement generating device disclosed in an embodiment of the present application;

[0047] Figure 3 This is a hardware structure block diagram of an advertisement generation device disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0049] An embodiment of the present application provides an advertisement generation method, which can be applied to various advertisement delivery systems or advertisement production systems, and can also be applied to various computer terminals or smart terminals. Its execution subject can be a processor or server of a computer terminal or smart terminal.

[0050] Next, combine Figure 1 The advertisement generation method of this application is described in detail, including the following steps:

[0051] Step S1: Obtain the target large model agent and the voice signal input by the user.

[0052] Specifically, the agent may include an intent recognition layer, a framework generation layer, a decision-making layer, and an editing layer.

[0053] The voice signal may be voice input by the user multiple times continuously, or may be voice input by the user once.

[0054] The user's multiple voice inputs can be integrated to determine the user's advertising strategy.

[0055] Step S2: Based on the intention recognition level, perform intent recognition on the voice signal to determine the user's advertising strategy.

[0056] Specifically, based on the intent recognition level, multi-dimensional intent recognition such as keyword extraction and voice analysis can be performed on the voice signal to determine the user's advertising construction style, advertising playback requirements, advertising objects, event development and / or advertising climax and other advertising presentation content and advertising delivery platforms and / or advertising delivery time and other advertising delivery strategies.

[0057] The advertisement object can be a physical product, a virtual product, or even any function of a virtual product, such as the matching function of a game.

[0058] The advertising construction style can include eye-catching appearance, conflicting plots, suspenseful blank spaces, exaggerated copywriting, humorous and funny, sweet plots and sad plots, etc.

[0059] Step S3: Generate a narrative framework based on the advertising strategy by using the framework to generate a hierarchy.

[0060] Specifically, a framework generation hierarchy built in combination with expert experience can be used to generate a narrative framework based on advertising strategies.

[0061] The narrative framework can cover the main presentation content of the four core stages of the complete advertising communication chain: beginning, development, climax and ending.

[0062] Next, this application will provide a specific advertising strategy and narrative framework example to illustrate this step.

[0063] The text message asks for a video ad for the TIT news feed featuring a couple. The video should feature a couple image as the hook for the first three seconds. After downloading TIT, the user is directed to enter their rank information to be matched with their partner. The user then plays a sweet duo with the person, followed by a highlight reel. Finally, a couple search sequence is played, accompanied by romantic background music.

[0064] The advertising strategy can be an eye-catching opening (sweet plot) + product introduction + game card point mix + drop version.

[0065] The narrative framework can be as follows:

[0066] Don't be envious, your CP is waiting for you in TIT! Sweet plot real-life plot playmate matching function

[0067] The video quickly shows several Valentine's Day limited character cards, and the pictures are full of romance.

[0068] Beginning: A girl with long pink hair is holding a mobile phone. She looks at the screen, tilts her head slightly and smiles, then closes her eyes, as if she is imagining that she can also have a sweet double-row experience.

[0069] Development: The girl clicks on the "campus section" of an app, which contains various study rooms. She seems to be looking for something, but soon the narration reminds her that if she wants to find a game partner, she should come and play.

[0070] Development: In the picture, a boy is wearing headphones and having a sweet voice chat with a girl. The avatar frames of the two people are floating on the screen, and they seem to be getting along very happily.

[0071] Climax: The two entered the game and started a fierce battle. Although the word "failure" appeared on the screen at the end, it can be heard from their conversation that they did not care about winning or losing, but enjoyed the process of playing together.

[0072] Epilogue: The social version of the game.

[0073] The narrative framework can be used to present the selling points of the advertising object, how to use the advertising object, and / or the pain points that the advertising object solves.

[0074] Step S4: utilizing the decision-making hierarchy and based on the narrative framework, obtaining matching constituent elements and constructing multiple storyboard scripts.

[0075] Specifically, the decision-making hierarchy can be used to divide the narrative framework according to each core stage, summarize each core stage, determine multiple storyboard overviews, and based on each storyboard overview, determine the material requirements, recall the matching components in multiple dimensions, and synthesize each component in stages to obtain multiple storyboard scripts.

[0076] Step S5: Based on the editing level, synthesize the storyboards to generate an advertisement.

[0077] Specifically, the editing level can be used to synthesize the various storyboards according to the development sequence corresponding to each core stage to obtain an advertisement.

[0078] Furthermore, different versions of advertisement costs may be synthesized for user selection, and the advertisement selected by the user may be combined into a film as an advertisement.

[0079] It can be seen from the above technical solutions that the advertisement generation method provided by the present application can obtain the voice signal input by the user, use the agent's intention recognition level to perform intent recognition on the voice signal, and determine the user's advertising strategy; based on this, the present application can interact with the user through the user's voice input, provide the user with a convenient way to input advertising needs, and enhance the user's participation in advertising editing and the convenience of advertising generation operations; determine the advertising strategy based on the voice input of each user, so that the advertising strategy is personalized and differentiated, thereby avoiding the homogenization of advertising content while making the advertisement more in line with user needs, improving the relevance of the advertising slogan to the target audience, and improving the advertising playback effect; on this basis, using the framework generation level, based on the advertising strategy, a narrative framework is generated; using the decision-making level, based on the narrative framework, the matching components are obtained to construct multiple storyboards; based on the editing level, the storyboards are synthesized to generate advertisements. Based on this, the present application can generate a narrative framework in accordance with the advertising strategy in a targeted manner, and generate multiple storyboards based on the narrative framework. The narrative framework can ensure the coherence and logic between the various storyboards, decompose the narrative framework into multiple storyboards, enrich the creative expression and form of the advertisement, greatly reduce the probability of advertisement homogeneity, and convert the entire advertisement generation process into a determination process and synthesis process of multiple storyboards, thereby simplifying the difficulty of advertisement generation; recalling the corresponding constituent elements of the narrative framework and then forming a storyboard can make the advertisement content fit the needs of different advertising stages, enhance the overall coordination and expressiveness of the advertisement, and enrich the content contained in the advertisement, thereby avoiding advertisement homogeneity. Advertising homogeneity; It can be seen that the target large model agent of this application has a clear division of labor at each level, and the voice interaction method and each level work closely together to provide users with convenient, efficient and personalized advertising generation services, improve user participation and satisfaction in the advertising production process, and generate advertisements based on user needs to ensure that the generated advertisements are unique and avoid stereotyped advertising content; process-based operations ensure efficient advertising production, and at the same time, professional-level processing improves the logic, coherence and overall quality of advertising content; users input voice signals to obtain advertisements that meet their needs, reduce dependence on manual design in the advertising generation process, reduce labor costs and time costs, and improve resource utilization efficiency.

[0080] Furthermore, the agent may also include a self-checking module.

[0081] The self-check module can be used to reflect on and revise target data and generate a modification plan;

[0082] and / or,

[0083] Respond to user adjustment instructions and generate modification plans. Adjustment instructions can also be voice input;

[0084] Utilizing the agent's intent recognition level, framework generation level, decision-making level, and editing level, target data is updated based on the modification plan, wherein the target data is the advertising strategy, the narrative framework, each storyboard, each constituent element, and / or the advertisement.

[0085] Furthermore, the self-check module is used to determine whether the correlation between each component and the matching storyboard overview exceeds a preset correlation threshold, and / or to determine whether each component of the same storyboard script matches, and / or to determine whether the text content in the same storyboard script is coherent. If not, a revision plan is generated.

[0086] The self-checking module can also score the advertisement, and when the score is lower than a threshold, the module re-enters step S2.

[0087] In some embodiments of the present application, the technical process of obtaining the target large model agent in step S1 is described in detail, and the steps are as follows:

[0088] S10. Obtain an initial agent model and different training advertisements, and perform strategy extraction on each training advertisement to obtain strategy data for each training advertisement. The initial agent model includes an initial intent recognition layer, an initial framework generation layer, an initial decision layer, and an initial editing layer.

[0089] Specifically, training advertisements may be obtained from different advertisement delivery platforms.

[0090] The advertising style, advertising object and advertising object introduction method of each training advertisement can be extracted, and the extracted information can be combined to obtain the strategy data of each training advertisement.

[0091] S11. Extract key features from the strategy data of each training advertisement to obtain multiple strategy subsets.

[0092] Specifically, key features can be extracted from the strategy data of each training advertisement to obtain multiple key fields corresponding to each strategy data;

[0093] Any multiple key fields corresponding to the same training advertisement are randomly combined to obtain multiple strategy subsets.

[0094] S12. Use the strategy data and each strategy subset corresponding to the same training advertisement as strategy samples.

[0095] Specifically, the strategy data corresponding to the same training advertisement and each strategy subset may be combined to form a strategy sample.

[0096] S13. Segment each training advertisement to obtain multiple advertisement segments.

[0097] Specifically, each training advertisement may be segmented according to the core phase to obtain multiple advertisement segments.

[0098] S14. Construct a training narrative framework for each training advertisement based on each advertisement segment of each training advertisement.

[0099] Specifically, the core stages of each training advertisement can be identified and summarized to generate a training narrative framework for each training advertisement.

[0100] S15. Extracting constituent elements from each advertising segment, determining the recall keywords corresponding to each constituent element, and determining the storyboard training script corresponding to each advertising segment.

[0101] Specifically, the relevant information such as character actions, event development trends, sound effects, lines and / or picture content of each advertising clip can be analyzed to extract the components of each advertising clip, and based on the training narrative framework and strategy samples, the recall keywords of each component can be determined. The components and recall keywords of each advertising clip can form a storyboard training script.

[0102] S16. Train the initial agent model according to the strategy sample, training narrative framework, and storyboard training script corresponding to each training advertisement until the initial agent model meets the preset stopping condition; and use the final initial agent model as the target large model agent.

[0103] Specifically, a plurality of speech training samples including matching strategy information are obtained, and the initial intent recognition level is trained using each speech training sample;

[0104] The initial frame generation layer, initial decision layer, and initial editing layer of the initial agent model are trained based on the strategy samples, training narrative framework, and storyboard training script corresponding to each training advertisement. The parameters of the initial frame generation layer, initial decision layer, and initial editing layer of the initial agent model are adjusted using the gradient descent method until the initial agent model converges. The final initial agent model is used as the target large model agent.

[0105] It can be seen from the above technical solution that this embodiment provides an optional method for training an agent. Through the above method, different existing advertisements can be integrated to train the agent, thereby accelerating the efficiency of advertisement generation.

[0106] In some embodiments of the present application, considering that the delivery effects of the same advertisement on different advertisement delivery platforms may be different, the present application attempts to add a process of fine-tuning the agent for the advertisement delivery platform. The following is a detailed description of the process, which includes the following steps:

[0107] S17. Obtain playback effects of different types of social advertisements on different advertising delivery platforms.

[0108] Specifically, you can obtain the conversion results, exposure volume, effective exposure ratio, coverage population, interaction rate, secondary dissemination rate, and stay time of different types of social ads on different advertising platforms.

[0109] S18. Obtain user feedback on different types of social ads on different advertising platforms.

[0110] Specifically, user feedback such as user advertising ratings, user comments, user messages, avoidance behaviors, and click behaviors of different types of social advertisements on different advertising delivery platforms can be obtained.

[0111] S19. Optimize parameters of the agent based on the playback effects and user feedback of different types of social advertisements on the same advertising delivery platform.

[0112] Specifically, the agent is parameter optimized based on user feedback and playback effects of different types of social advertisements on the same advertisement delivery platform, and user feedback and playback effects of the same social advertisement on different advertisement delivery platforms.

[0113] As can be seen from the above technical solution, compared to the previous embodiment, this embodiment adds an optional method for optimizing the agent. This method integrates the content creation, platform adaptation, and delivery strategies involved in advertising, allowing the ad generation process to take into account the user characteristics and dissemination patterns of different advertising delivery platforms, thereby improving the display effect of generated ads and continuously improving ad quality. By continuously optimizing the agent based on user feedback and playback effects, the effectiveness of agent-generated ads in attracting user attention, conveying core values, and driving purchasing behavior can be continuously improved.

[0114] In some embodiments of the present application, step S2, performing intent recognition on the voice signal based on the intent recognition level and determining the user's advertising strategy, is described in detail. The steps are as follows:

[0115] S20. Based on the intention recognition level, the voice signal is converted into text to obtain text information, and keywords are extracted from the text information to determine key fields containing advertising style and advertising content planning; combined with a preset advertising structure formula library, an advertising strategy matching the key fields is generated.

[0116] Specifically, the intent recognition layer can be used to convert the voice signal into text to obtain text information, and key fields such as advertising style and advertising planning can be extracted from the text information.

[0117] Based on the correspondence between each strategy subset and strategy data, and in combination with the expert experience database, the data types required for different types of advertising strategies can be determined;

[0118] Combined with the expert experience database, each strategy subset and strategy data is updated;

[0119] Build an advertising structure formula library that includes strategy data and the corresponding relationship between strategy subsets;

[0120] Selecting a strategy subset corresponding to the key field from the advertisement structure formula library, and determining matching strategy data based on the correspondence between the strategy subset and the strategy data;

[0121] The key fields are supplemented and optimized based on the data types contained in the matched strategy data to obtain the advertising strategy.

[0122] It can be seen from the above technical solution that this embodiment provides an optional method for performing intent recognition on the voice signal based on the intent recognition level and determining the user's advertising strategy. The above method can further improve the effectiveness of the advertising strategy.

[0123] In some embodiments of the present application, the process of step S4, using the decision hierarchy to obtain matching constituent elements based on the narrative framework and constructing multiple storyboards, is described in detail. The steps are as follows:

[0124] S40. Utilize the decision-making hierarchy and combine it with the advertising strategy to decompose the narrative framework and obtain multiple storyboard overviews. Make content decisions for each storyboard overview, determine the picture requirements, lens language, sound effect requirements, and storyboard emotions, and generate a storyboard script.

[0125] Specifically, each narrative framework can be disassembled according to each core stage, and the narrative content corresponding to each core stage can be briefly summarized to obtain a storyboard description;

[0126] You can make content decisions for each storyboard overview, determine the corresponding picture requirements, lens language, sound effect requirements and storyboard emotions for each storyboard, and obtain the storyboard script.

[0127] It can be seen from the above technical solution that this embodiment provides an optional method of utilizing the decision-making hierarchy, based on the narrative framework, to obtain matching constituent elements and construct multiple storyboard scripts. Through the above method, the storyboard script can be composed of factors such as picture content, lens language, sound effects, storyboard emotions and text content, thereby enriching the advertising content and improving the quality of advertising.

[0128] In some embodiments of the present application, the process of making content decisions for each storyboard overview, determining the image requirements, shot language, sound effect requirements, and storyboard emotions, and generating a storyboard script in step S40 is described in detail, and the steps are as follows:

[0129] S400. Utilize the decision-making hierarchy to make content decisions for each storyboard overview, determine the picture requirements, lens language, sound effect requirements, and storyboard emotions, and perform multi-dimensional recall of video materials based on the picture requirements and lens language corresponding to each storyboard overview, obtain storyboard videos that match the picture requirements and lens language, and select matching storyboard sound effects and storyboard timbre based on the sound effect requirements and storyboard emotions corresponding to each storyboard overview. Generate storyboard dialogues based on each storyboard overview and its corresponding picture requirements, lens language, and storyboard emotions; the storyboard videos, storyboard sound effects, storyboard timbre, and storyboard dialogues of the same storyboard overview constitute a corresponding storyboard script.

[0130] Specifically, the decision-making hierarchy is used to generate semantic vectors and storyboard keywords based on the overview of each storyboard, and the object subject, action description, scene atmosphere, etc. are determined.

[0131] Screen requirements may include semantic vectors, storyboard keywords, advertising objects, action descriptions, and scene atmosphere;

[0132] Camera language includes zoom, long shot, close-up and / or close-up.

[0133] Sound effects requirements can be cheerful, happy, funny, exciting and / or heroic, etc.

[0134] The emotions of the storyboards can be to convey intimacy, reassurance, comfort, create a sense of oppression and / or enthusiasm, etc.

[0135] The mood of the storyboard can be matched to the sound requirements.

[0136] After recalling each video based on the picture requirements and lens language, the semantic vector and storyboard keywords can be used to sort the videos and select the storyboard video that best meets the picture requirements.

[0137] Call various tool methods, and select the most matching storyboard sound effects and storyboard timbres based on the comprehensive sound effect requirements and storyboard emotions.

[0138] Combining each storyboard overview and its corresponding screen content, lens language, and storyboard emotions, one or more storyboard lines corresponding to each storyboard overview are generated.

[0139] The storyboard videos, storyboard sound effects, storyboard timbres and storyboard lines of the same storyboard overview can be combined to form a corresponding storyboard script.

[0140] It can be seen from the above technical solution that this embodiment provides an optional method for generating a storyboard script by integrating picture requirements, lens language, sound effect requirements and storyboard emotions. Through the above method, multi-dimensional content can be integrated to generate a storyboard script, thereby improving the appeal and executability of the storyboard script.

[0141] In some embodiments of the present application, step S5, the process of synthesizing the storyboards based on the editing level to generate an advertisement is described in detail, and the steps are as follows:

[0142] S50. Utilizing the editing level, for each storyboard script, with the goal of maximizing the degree of match between the storyboard overview of the storyboard script and each constituent element, adjust the image content within the storyboard video of the storyboard script, adjust the attributes of the storyboard timbre and storyboard sound effects of the storyboard script, optimize the storyboard lines of the storyboard script, and present the optimized storyboard lines with the adjusted storyboard timbre to obtain storyboard voice, and perform structured synthesis on the adjusted storyboard video, storyboard sound effects, and storyboard voice to obtain an advertisement.

[0143] Specifically, the editing level can be used to optimize the picture content of the storyboard video for each storyboard script according to the picture requirements and lens language, including but not limited to adding a pattern cursor, enlarging the video content, and changing the character image.

[0144] According to the degree of connection between each storyboard script, the tone and frequency of each storyboard sound effect are adjusted.

[0145] According to the degree of connection between different storyboard lines in each storyboard script, each storyboard line is optimized and adjusted, and the optimized storyboard lines are performed with the adjusted storyboard timbre to obtain storyboard voices of different storyboard scripts, and the adjusted storyboard video, storyboard sound effect and storyboard voice are structured and synthesized to obtain an advertisement.

[0146] It can be seen from the above technical solution that this embodiment provides an optional method for synthesizing advertisements. Through the above method, this application can improve the coherence of the entire advertisement by adjusting the constituent elements of each storyboard script, thereby improving the rendering ability and logic of the advertisement.

[0147] Next, we will combine Figure 2The advertisement generation device provided in this application is introduced in detail. The advertisement generation device provided below can be compared with the advertisement generation method provided above.

[0148] See also Figure 2 It can be found that the advertisement generating means may include:

[0149] Acquisition module 10, used to acquire the target large model agent and the voice signal input by the user, the agent includes the intention recognition layer, the framework generation layer, the decision layer and the editing layer;

[0150] a determination module 20, configured to perform intent recognition on the voice signal based on the intent recognition level and determine the user's advertising strategy;

[0151] A generating module 30 for generating a narrative framework based on the advertising strategy by using the framework generation hierarchy;

[0152] A construction module 40 is configured to utilize the decision-making hierarchy and the narrative framework to obtain matching constituent elements and construct a plurality of storyboards;

[0153] The synthesis module 50 is used to synthesize the various storyboards based on the editing level to generate an advertisement.

[0154] Furthermore, the acquisition module 10 may include:

[0155] a first acquisition unit, configured to acquire an initial agent model and different training advertisements, and perform strategy extraction on each training advertisement to obtain strategy data for each training advertisement, wherein the initial agent model includes an initial intent recognition layer, an initial framework generation layer, an initial decision layer, and an initial editing layer;

[0156] A second acquisition unit is used to extract key features from the strategy data of each training advertisement to obtain multiple strategy subsets;

[0157] a third acquisition unit, configured to use the strategy data and each strategy subset corresponding to the same training advertisement as strategy samples;

[0158] a fourth acquisition unit, configured to segment each training advertisement to obtain a plurality of advertisement segments;

[0159] a fifth acquisition unit, configured to construct a training narrative framework for each training advertisement based on each advertisement segment of each training advertisement;

[0160] a sixth acquisition unit, configured to extract components from each advertisement segment, determine recall keywords corresponding to each component, and determine a storyboard training script corresponding to each advertisement segment;

[0161] The seventh acquisition unit is used to train the initial agent model according to the strategy sample, training narrative framework and storyboard training script corresponding to each training advertisement until the initial agent model meets the preset stopping condition; and the final initial agent model is used as the target large model agent.

[0162] Furthermore, the acquisition module 10 may further include:

[0163] An eighth obtaining unit, configured to obtain playback effects of different types of social advertisements on different advertising delivery platforms;

[0164] A ninth acquisition unit, configured to acquire user feedback on different types of social advertisements on different advertising delivery platforms;

[0165] The tenth acquisition unit is used to optimize the parameters of the agent based on the playback effects and user feedback of different types of social advertisements on the same advertising delivery platform.

[0166] Furthermore, the determination module 20 may include:

[0167] a first determining unit configured to convert the voice signal into text based on the intent recognition level to obtain text information, and extract keywords from the text information to determine key fields containing advertisement style and advertisement content planning;

[0168] The second determining unit is configured to generate an advertising strategy matching the key fields in combination with a preset advertising structure formula library.

[0169] Furthermore, the building block 40 may include:

[0170] The storyboard script generation unit is used to utilize the decision-making hierarchy and combine it with the advertising strategy to disassemble the narrative framework, obtain multiple storyboard overviews, make content decisions for each storyboard overview, determine the picture requirements, lens language, sound effect requirements and storyboard emotions, and generate a storyboard script.

[0171] Furthermore, the storyboard generation unit may include:

[0172] The constituent element determination subunit is used to make content decisions for each storyboard overview by utilizing the decision-making hierarchy, determine the picture requirements, lens language, sound effect requirements and storyboard emotions, and perform multi-dimensional recall of video materials based on the picture requirements and lens language corresponding to each storyboard overview, obtain storyboard videos that match the picture requirements and lens language, and select matching storyboard sound effects and storyboard timbre based on the sound effect requirements and storyboard emotions corresponding to each storyboard overview, and generate storyboard lines based on each storyboard overview and its corresponding picture requirements, lens language, and storyboard emotions; the storyboard videos, storyboard sound effects, storyboard timbre and storyboard lines of the same storyboard overview constitute the corresponding storyboard script.

[0173] Furthermore, the synthesis module 50 may include:

[0174] The storyboard script synthesis unit is used to utilize the editing level to, for each storyboard script, adjust the picture content within the storyboard video of the storyboard script, adjust the attributes of the storyboard timbre and storyboard sound effects of the storyboard script, optimize the storyboard lines of the storyboard script, present the optimized storyboard lines with the adjusted storyboard timbre, obtain storyboard voice, and perform structured synthesis of the adjusted storyboard video, storyboard sound effects and storyboard voice to obtain an advertisement.

[0175] The advertisement generation device provided in the embodiment of the present application can be applied to advertisement generation devices, such as PC terminals, cloud platforms, servers and server clusters. Figure 3 The hardware structure diagram of the advertisement generating device is shown. Figure 3 ,The hardware structure of the advertisement generating device may include: at least one processor 1, at least one communication interface 2, at least one memory 3 and at least one communication bus 4;

[0176] In the embodiment of the present application, the number of the processor 1, the communication interface 2, the memory 3, and the communication bus 4 is at least one, and the processor 1, the communication interface 2, and the memory 3 communicate with each other through the communication bus 4;

[0177] The processor 1 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention;

[0178] The memory 3 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory;

[0179] The memory stores a program, and the processor can call the program stored in the memory, wherein the program is used to:

[0180] Obtain the target large model agent and the user's input voice signal, wherein the agent includes an intent recognition layer, a framework generation layer, a decision layer, and an editing layer;

[0181] Based on the intention recognition level, performing intent recognition on the voice signal to determine the user's advertising strategy;

[0182] generating a narrative framework based on the advertising strategy using the framework generation hierarchy;

[0183] Utilizing the decision-making hierarchy and based on the narrative framework, obtaining matching constituent elements and constructing multiple storyboards;

[0184] Based on the editing level, each storyboard script is synthesized to generate an advertisement.

[0185] Optionally, the refined functions and extended functions of the program may refer to the above description.

[0186] The present application also provides a readable storage medium, which may store a program suitable for execution by a processor, wherein the program is used to:

[0187] Obtain the target large model agent and the user's input voice signal, wherein the agent includes an intent recognition layer, a framework generation layer, a decision layer, and an editing layer;

[0188] Based on the intention recognition level, performing intent recognition on the voice signal to determine the user's advertising strategy;

[0189] generating a narrative framework based on the advertising strategy using the framework generation hierarchy;

[0190] Utilizing the decision-making hierarchy and based on the narrative framework, obtaining matching constituent elements and constructing multiple storyboards;

[0191] Based on the editing level, each storyboard script is synthesized to generate an advertisement.

[0192] Optionally, the refined functions and extended functions of the program may refer to the above description.

[0193] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0194] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0195] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. The various embodiments of the present application may be combined with each other. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for generating an advertisement, characterized in that: include: Obtain the target large model agent and the user's input voice signal, wherein the agent includes an intent recognition layer, a framework generation layer, a decision layer, and an editing layer; Based on the intention recognition level, performing intent recognition on the voice signal to determine the user's advertising strategy; generating a narrative framework based on the advertising strategy using the framework generation hierarchy; Utilizing the decision-making hierarchy and based on the narrative framework, obtaining matching constituent elements and constructing multiple storyboards; Based on the editing level, each storyboard script is synthesized to generate an advertisement.

2. The advertisement generating method according to claim 1, characterized in that: The step of obtaining the target large model agent includes: Obtaining an initial agent model and different training ads, and performing strategy extraction on each training ad to obtain strategy data for each training ad, wherein the initial agent model includes an initial intent recognition layer, an initial framework generation layer, an initial decision layer, and an initial editing layer; Extract key features from the strategy data of each training ad to obtain multiple strategy subsets; The strategy data and each strategy subset corresponding to the same training advertisement are used as strategy samples; Segment each training advertisement to obtain multiple advertisement segments; Based on each advertising segment of each training advertisement, a training narrative framework of each training advertisement is constructed; Extract the components from each advertising clip, determine the recall keywords corresponding to each component, and determine the storyboard training script corresponding to each advertising clip; The initial agent model is trained according to the strategy sample, training narrative framework and storyboard training script corresponding to each training advertisement until the initial agent model meets the preset stopping condition; the final initial agent model is used as the target large model agent.

3. The advertisement generating method according to claim 2, characterized in that: Also includes: Obtain the playback effects of different types of social ads on different advertising platforms; Obtain user feedback on different types of social ads on different advertising platforms; Based on the playback effects and user feedback of different types of social advertisements on the same advertising platform, the parameters of the agent are optimized.

4. The advertisement generating method according to claim 1, wherein: The performing intent recognition on the voice signal based on the intent recognition level and determining the user's advertising strategy includes: Based on the intent recognition level, convert the voice signal into text to obtain text information, and extract keywords from the text information to determine key fields containing advertising style and advertising content planning; Combined with the preset advertisement structure formula library, an advertisement strategy matching the key fields is generated.

5. The advertisement generating method according to claim 1, wherein: The method utilizes the decision-making hierarchy and, based on the narrative framework, obtains matching constituent elements and constructs multiple storyboards, including: By utilizing the decision-making hierarchy and combining it with the advertising strategy, the narrative framework is disassembled to obtain multiple storyboard overviews, and content decisions are made for each storyboard overview to determine the picture requirements, lens language, sound effect requirements and storyboard emotions, and generate a storyboard script.

6. The advertisement generating method according to claim 5, characterized in that: The above process involves making content decisions for each storyboard overview, determining the image requirements, lens language, sound effect requirements, and storyboard emotions, and generating a storyboard script, including: The decision-making hierarchy is used to make content decisions for each storyboard overview, determine the picture requirements, lens language, sound effect requirements and storyboard emotions, and perform multi-dimensional recall of video materials based on the picture requirements and lens language corresponding to each storyboard overview, obtain storyboard videos that match the picture requirements and lens language, and select matching storyboard sound effects and storyboard timbre based on the sound effect requirements and storyboard emotions corresponding to each storyboard overview, and generate storyboard lines based on each storyboard overview and its corresponding picture requirements, lens language, and storyboard emotions; the storyboard videos, storyboard sound effects, storyboard timbre and storyboard lines of the same storyboard overview constitute the corresponding storyboard script.

7. The advertisement generating method according to claim 6, characterized in that: The step of synthesizing the storyboards based on the editing level to generate an advertisement includes: By utilizing the editing level, for each storyboard script, with the goal of maximizing the degree of match between the storyboard overview of the storyboard script and each constituent element, the screen content within the storyboard video of the storyboard script is adjusted, the storyboard timbre and storyboard sound effects of the storyboard script are attribute adjusted, the storyboard lines of the storyboard script are optimized, and the optimized storyboard lines are presented with the adjusted storyboard timbre to obtain storyboard voice, and the adjusted storyboard video, storyboard sound effects and storyboard voice are structured and synthesized to obtain an advertisement.

8. An advertisement generating device, characterized in that: include: The acquisition module is used to obtain the target large model agent and the voice signal input by the user. The agent includes the intention recognition layer, the framework generation layer, the decision layer and the editing layer; a determination module, configured to perform intent recognition on the voice signal based on the intent recognition level and determine the user's advertising strategy; a generating module for generating a narrative framework based on the advertising strategy by using the framework to generate a hierarchy; A construction module, configured to utilize the decision-making hierarchy and, based on the narrative framework, obtain matching constituent elements and construct a plurality of storyboards; A synthesis module is used to synthesize various storyboards based on the editing level to generate advertisements.

9. An advertisement generating device, characterized in that: including memory and processor; The memory is used to store programs; The processor is configured to execute the program to implement each step of the advertisement generation method according to any one of claims 1 to 7.

10. A readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, each step of the advertisement generation method according to any one of claims 1 to 7 is implemented.

Images