Video script generation method and device, computer equipment and storage medium

Abstract

The embodiment of the invention provides a video script generation method and device, computer equipment and a storage medium, and relates to the technical field of artificial intelligence. The method comprises the following steps: acquiring script demand description information; acquiring script framework information according to the script demand description information; utilizing the large language model and the first cue word to generate a plurality of script segments based on a pre-constructed video script knowledge base and script framework information, and generating an initial video script according to the plurality of script segments; the first cue word comprises a script segment cue word for each script segment in the multiple script segments; generating a target video script based on the initial video script by using the large language model and the second cue word; the second cue word is used for guiding the large language model to adjust the script data format of the initial video script. According to the method, the efficiency and accuracy of video script generation are improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, apparatus, computer equipment, storage medium and computer program product for generating a video script. Background Art

[0002] In the field of video shooting, it is usually necessary to generate the corresponding video script in advance. Given that the traditional method that relies on manual creation is inefficient and easily limited by the creator's personal experience and style, in general, in video script creation technology, a large language model can be introduced to generate video scripts using the large language model, which can improve the efficiency of video script generation.

[0003] However, in the current video script generation technology, the video information obtained during the information processing process is relatively single and insufficient. For example, basic information is extracted from the video, such as lines and pictures, while the potential cultural connotations and emotional expressions are ignored, resulting in the generated video script being significantly insufficient in logical coherence, cultural adaptability and emotional resonance, and having problems of low accuracy and low practicality. Summary of the invention

[0004] Based on this, it is necessary to provide a video script generation method, device, computer equipment, storage medium and computer program product to address the above technical problems.

[0005] In a first aspect, the present application provides a method for generating a video script. The method comprises:

[0006] Get the script requirement description information;

[0007] According to the script requirement description information, obtain script framework information;

[0008] Using the large language model and the first prompt word, based on the pre-built video script knowledge base and the script framework information, multiple script segments are generated in segments, and an initial video script is generated according to the multiple script segments; the first prompt word includes a script segment prompt word for each of the multiple script segments;

[0009] The target video script is generated based on the initial video script by using the large language model and the second prompt word; the second prompt word is used to guide the large language model to adjust the script data format of the initial video script.

[0010] In one of the embodiments, before the use of the large language model and the first prompt word to generate multiple script segments in segments based on a pre-built video script knowledge base and the script framework information, it also includes: obtaining multiple historical video scripts in a historical time period corresponding to the current time period; obtaining multiple script elements based on the historical video script information of each of the multiple historical video scripts; obtaining element attribute information of each of the script elements, and obtaining element relationships between each of the script elements based on the element attribute information; and constructing the video script knowledge base based on the multiple script elements and the element relationships between each of the script elements.

[0011] In one of the embodiments, the use of a large language model and a first prompt word, based on a pre-built video script knowledge base and the script framework information, to generate multiple script segments in segments, includes: determining a script structure according to the script framework information; screening out a matching video script that matches the script framework information from multiple historical video scripts according to the script framework information and multiple script elements of each of the multiple historical video scripts contained in the video script knowledge base and the element relationship between the script elements; and using the large language model and the first prompt word to generate multiple script segments in segments according to the script structure and the matching video script.

[0012] In one of the embodiments, the method of screening out a matching video script that matches the script framework information from a plurality of historical video scripts based on the script framework information and a plurality of script elements of each of the plurality of historical video scripts contained in the video script knowledge base and the element relationships between the script elements comprises: obtaining the similarity between the script framework information and each of the plurality of historical video scripts based on the script framework information and a plurality of script elements of each of the plurality of historical video scripts contained in the video script knowledge base and the element relationships between the script elements; and determining the historical video script having the similarity greater than a preset similarity threshold as the matching video script.

[0013] In one of the embodiments, obtaining script framework information according to the script requirement description information includes: obtaining associated script keywords according to the script requirement description information; obtaining a target associated video in an associated video data source according to the associated script keywords; obtaining multidimensional features of the target associated video according to video information of the target associated video; the video information includes at least two types of text information, audio information and video image information of the target associated video; the multidimensional features include at least emotional expression features; and obtaining the script framework information according to the multidimensional features.

[0014] In one of the embodiments, obtaining a target associated video in an associated video data source according to the associated script keywords includes: obtaining multiple candidate associated videos in an associated video data source according to the associated script keywords; obtaining script requirement characteristics according to the script requirement description information; and screening out the target associated video from the multiple candidate associated videos according to the script requirement characteristics.

[0015] In one of the embodiments, before generating a target video script based on the initial video script using the large language model and the second prompt word, the method further includes: determining initial data format information of the initial video script; determining target data format information of the target video script; and generating the second prompt word based on the initial data format information and the target data format information.

[0016] In a second aspect, the present application provides a video script generation device. The device comprises:

[0017] Information acquisition module, used to obtain script requirement description information;

[0018] A framework acquisition module, used to acquire script framework information according to the script requirement description information;

[0019] A first script generation module is used to segment a plurality of script segments based on a pre-built video script knowledge base and the script framework information using a large language model and a first prompt word, and generate an initial video script according to the plurality of script segments; the first prompt word includes a script segment prompt word for each of the plurality of script segments;

[0020] The second script generation module is used to generate a target video script based on the initial video script by using the large language model and the second prompt word; the second prompt word is used to guide the large language model to adjust the script data format of the initial video script.

[0021] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0022] Get the script requirement description information;

[0023] According to the script requirement description information, obtain script framework information;

[0024] Using the large language model and the first prompt word, based on the pre-built video script knowledge base and the script framework information, multiple script segments are generated in segments, and an initial video script is generated according to the multiple script segments; the first prompt word includes a script segment prompt word for each of the multiple script segments;

[0025] The target video script is generated based on the initial video script by using the large language model and the second prompt word; the second prompt word is used to guide the large language model to adjust the script data format of the initial video script.

[0026] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0027] Get the script requirement description information;

[0028] According to the script requirement description information, obtain script framework information;

[0029] Using the large language model and the first prompt word, based on the pre-built video script knowledge base and the script framework information, multiple script segments are generated in segments, and an initial video script is generated according to the multiple script segments; the first prompt word includes a script segment prompt word for each of the multiple script segments;

[0030] The target video script is generated based on the initial video script by using the large language model and the second prompt word; the second prompt word is used to guide the large language model to adjust the script data format of the initial video script.

[0031] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0032] Get the script requirement description information;

[0033] According to the script requirement description information, obtain script framework information;

[0034] Using the large language model and the first prompt word, based on the pre-built video script knowledge base and the script framework information, multiple script segments are generated in segments, and an initial video script is generated according to the multiple script segments; the first prompt word includes a script segment prompt word for each of the multiple script segments;

[0035] The target video script is generated based on the initial video script by using the large language model and the second prompt word; the second prompt word is used to guide the large language model to adjust the script data format of the initial video script.

[0036] In the above-mentioned video script generation method, device, computer equipment, storage medium and computer program product, first, script requirement description information can be obtained; next, script framework information can be obtained according to the script requirement description information; further, a large language model and a first prompt word can be used to generate multiple script segments in segments based on a pre-built video script knowledge base and script framework information, and an initial video script is generated according to multiple script segments; the first prompt word includes a script segment prompt word for each script segment in multiple script segments; finally, a target video script can be generated based on the initial video script using a large language model and a second prompt word; the second prompt word is used to guide the large language model to adjust the script data format of the initial video script. In the method provided by the embodiment of the present application, script framework information can be obtained according to the script requirement description information and in combination with hot spot information, and a large language model and a pre-built video script knowledge base can be used to generate multiple script segments in segments based on the script framework information, and then an initial video script can be generated, which can improve the efficiency and accuracy of video script generation, and subsequently, the initial video script can be optimized to obtain a target video script, which can further improve the accuracy of video script generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 A schematic diagram of a flow chart of a method for generating a video script provided in an embodiment of the present application;

[0039] Figure 2 A schematic diagram of a process for obtaining script framework information provided in an embodiment of the present application;

[0040] Figure 3 A schematic diagram of a process for building a video script knowledge base provided in an embodiment of the present application;

[0041] Figure 4 A schematic diagram of a process for generating multiple script segments in segments provided in an embodiment of the present application;

[0042] Figure 5A schematic diagram of a process for obtaining multi-dimensional features of a target-related video provided in an embodiment of the present application;

[0043] Figure 6 A schematic diagram of a process for obtaining target-related videos provided in an embodiment of the present application;

[0044] Figure 7 A schematic diagram of a process for generating a second prompt word provided in an embodiment of the present application;

[0045] Figure 8 A structural block diagram of a video script generating device provided in an embodiment of the present application;

[0046] Fig. 9 An internal structure diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0048] In an exemplary embodiment, Figure 1 As shown, a method for generating a video script is provided. This embodiment uses the method applied to a server as an example for illustration. It is understandable that the method can also be applied to a terminal, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0049] Step 102, obtaining script requirement description information.

[0050] Among them, the video script generation method of the embodiment of the present application can be applied to a video script generation system, and the video script generation system may include a video script generation terminal and a system server. The video script generation terminal may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, Internet of Things devices and portable wearable devices, and the Internet of Things devices may be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. Portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The system server may be a server for generating video scripts in the video script generation system, and the system server may be implemented with an independent server or a server cluster composed of multiple servers. The video script generation terminal communicates with the system server through a network. The video script generation personnel can register and log in to the video script generation system on the video script generation terminal, and the video script generation terminal responds to the login operation triggered by the video script generation personnel and displays a video script requirement input page; the video script generation personnel can enter the script requirement description information on the video script requirement input page to trigger a video script generation request for the video script to be generated, and the video script generation request includes the script requirement description information. In a possible implementation, the video script to be generated may be a video promotion script for a target object, for example, a promotional video of a target product, and the script requirement description information may be a promotion requirement for a target object, for example, a promotional video of a target product. The script requirement description information may include the requirement information for the promotion of the target product and the requirement information for the creation of the video script, for example, it may include basic information, creative references, product characteristics, picture and language style, role requirements, plot and scene design, format specifications, audience orientation, and special requirements, and the script requirement description information may be used as guidance information for generating the video script to be generated. Among them, the basic information may include the video subject, video direction or core concept (for example, technology product display and cultural activity promotion, etc.), video purpose (for example, advertising, education and teaching, and entertainment programs, etc.) and target audience of the video to be generated. The characteristics of the target audience may include age range (for example, children, teenagers, adults, and the elderly, etc.), hobbies (for example, technology enthusiasts, art seekers, and sports enthusiasts, etc.), and cultural background, etc. The creative reference source may include a designated reference video link or description and a hot trend or popular element. The product or service characteristics may include product core advantages, brand image, and values, etc.The requirements for picture and visual style may include color preferences (e.g., warm tones, cool tones, bright colors, and retro colors), scene setting preferences (e.g., natural scenery scenes (forests, beaches, mountains, etc.), urban environment scenes (streets, high-rise buildings, parks, etc.), indoor scenes (homes, offices, studios, etc.), and scene style characteristics (such as modern simplicity, classical elegance, sci-fi futuristic, etc.)) and lens language requirements (e.g., preferences or special requirements for lens movement (push, pull, shake, move, follow, etc.), shooting angles (overhead, overhead, level, etc.), and scene types (long shot, panoramic, mid-shot, close-up, close-up, etc.)). The requirements for language style and lines may include the overall language style (e.g., humorous, serious, professional, warm and touching, and passionate, etc.) and line characteristics. The requirements for roles and character images may include the main role characteristics (e.g., personality characteristics of the protagonist and supporting roles (cheerful, introverted, brave, wise, etc.), appearance characteristics (age, body shape, appearance description, etc.), role background (occupation, growth experience, etc.)) and role relationship settings (e.g., friends, family, etc.). The plot and story structure requirements may include plot clues and development direction (for example, the setting of the beginning, development, climax and ending of the story, as well as the turning points of the plot, suspense settings, etc.), and story structure preferences (for example, linear narrative, non-linear narrative (flashback, interpolation and multi-line narrative, etc.) and circular structure, etc.). The script format and specification requirements may include script format examples and duration requirements. The audience interaction and guidance requirements may include emotional resonance and interaction points, etc., which may be the emotional resonance points generated with the audience in the video (for example, touching, inspiring and enlightening, etc.), as well as the need to set up interactive links (for example, asking questions, guiding comments and likes, etc.).

[0051] Step 104, obtaining script framework information according to the script requirement description information.

[0052] Among them, Figure 2As shown, first, based on the script requirement description information, it can be determined whether the script requirement description information contains video direction information; if the script requirement description information contains the video direction information, it can be determined whether the script requirement description information contains a reference video link; if the script requirement description information contains the reference video link, the reference video link is subjected to video parsing to obtain script framework information for generating the video script to be generated; if the script requirement description information does not contain the reference video link, the associated script keywords of the video script to be generated are extracted based on the script requirement description information. Next, based on the associated script keywords, multiple candidate associated videos corresponding to the video script to be generated can be obtained from the script requirement description information or the associated video data source of the video script to be generated. Furthermore, a target associated video with the highest relevance to the video script to be generated can be screened out from the multiple candidate associated videos, and video parsing can be performed on the target associated video to obtain script framework information for generating the video script to be generated. If the script requirement description information does not contain the video direction information, it is determined whether the script requirement description information contains hotspot related information. If the script requirement description information If the information contains hotspot related information, then the associated real-time hotspot corresponding to the script requirement description information is obtained from the associated hotspot data source, and the associated hotspot words are filtered out according to the associated real-time hotspot. Next, based on the associated hotspot words, multiple candidate associated videos corresponding to the video script to be generated can be obtained from the script requirement description information or the associated video data source of the video script to be generated. Further, the target associated video with the highest relevance to the video script to be generated can be filtered out from the multiple candidate associated videos, and the video of the target associated video can be analyzed to obtain the script framework information for generating the video script to be generated. If the script requirement description information does not contain hotspot related information, then the corresponding script keywords are determined according to the script requirement description information. Next, based on the script keywords, multiple candidate associated videos corresponding to the video script to be generated can be obtained from the script requirement description information or the associated video data source of the video script to be generated. Further, the target associated video with the highest relevance to the video script to be generated can be filtered out from the multiple candidate associated videos, and the video of the target associated video can be analyzed to obtain the script framework information for generating the video script to be generated.

[0053] Step 106, using the large language model and the first prompt word, based on the pre-built video script knowledge base and script framework information, segmentally generate multiple script segments, and generate an initial video script based on the multiple script segments.

[0054] In this step, a video script knowledge base can be pre-constructed, and the video script knowledge base contains multiple historical video scripts. The construction process of the video script knowledge base may include: obtaining multiple historical video scripts in the historical time period corresponding to the current time period from multiple video script sources (for example, corresponding business systems, or pre-written script sets, etc.); obtaining multiple script elements based on the historical video script information of each historical video script in the multiple historical video scripts; obtaining element attribute information of each script element, and obtaining the element relationship between each script element based on the element attribute information; constructing a video script knowledge base based on multiple script elements and the element relationship between each script element. Furthermore, a large language model and a first prompt word can be used to generate multiple script segments in segments based on the pre-constructed video script knowledge base and script framework information, and an initial video script is generated based on the multiple script segments. Among them, the first prompt word of the large language model can be pre-set, and the first prompt word includes a script segment prompt word for each script segment in multiple script segments. For example, the first prompt word can be set as "You are an experienced information flow advertising script copywriting expert, good at extracting valuable data from various sources and integrating it into an easy-to-understand form. You have a deep understanding of video content analysis and can accurately grasp the audience's interests and behavior patterns. You will receive a message: reference video creative content: for reference to the character lines / copywriting / creativity, etc.". Next, the video script knowledge base can be screened for matching video scripts with the script requirement description information; further, the large language model can be used to parse the script framework information to obtain a script structure for generating the video script to be generated, and the script structure can include the video architecture and key elements of the video script to be generated, such as video direction, video subject or core concept, target audience, emotional tone and visual elements. Finally, the matching video script can be divided into multiple matching script segments, for example, it can be divided into an introduction matching script segment, a main body matching script segment and an ending matching script segment. Then, a large language model can be used, combined with the script structure and the multiple matching script segments, to generate multiple script segments by segmented imitation, and the initial video script can be generated based on the multiple script segments.

[0055] Step 108, using the large language model and the second prompt word, based on the initial video script, generate a target video script.

[0056] Among them, the second prompt word is used to guide the large language model to adjust the script data format of the initial video script. The script data format may include a script logical sequence and a data structure, and the script data format of the video script to be generated may be set in advance in the second prompt word. The initial video script is input into the large language model. Under the guidance of the second prompt word, the large language model may check and optimize the logical sequence of the initial video script. For example, the logical sequence of the initial video script may be to first determine the theme and core plot, and then gradually expand the role shaping, scene design, and line writing, and finally perform an overall logic check and optimization. In addition, the large language model may also convert the data structure of the initial video script from unstructured data to structured data, and then the converted structured data may be verified and cleaned to ensure the accuracy and integrity of the data. In addition, in this step, the organized structured data can also be combined with visualization tools to realize the visualization presentation of the script content. For example, using graphical script editing software, according to the scene description, lens view and other information in the structured data, a corresponding graphical representation is generated, such as a lens sequence on the timeline, a scene layout diagram, etc., so that the script creator and related personnel can intuitively see the overall structure and visual effect of the script. At the same time, according to the attributes such as the emotional atmosphere in the data, corresponding colors or marks can be added to the visualization elements to enhance the visual experience. And realize human-computer interactive modification, the video script generator can modify the script directly on the video script generation terminal based on the visualization presentation. The user can adjust the order by dragging the lens on the timeline, click on the scene element to edit (for example, modify the screen content, lines, etc.), and select different lens view or emotional atmosphere options to change. When the user makes a modification, the system updates the modified information to the structured data in real time and feeds it back to the large language model. The model provides further optimization suggestions or automatically adjusts the relevant content based on the new input and modification intention, realizing efficient collaboration and iterative optimization between man and machine. For example, when a user shortens the duration of a scene, the system automatically updates the duration field in the structured data and asks the large language model whether the rhythm of the lines or the content of the picture needs to be adjusted when the duration is shortened. The model provides corresponding answers based on its knowledge and experience, and the user can continue to make modifications based on the answers.

[0057] In the method of this embodiment, first, the script requirement description information can be obtained; next, the script framework information can be obtained according to the script requirement description information; further, the large language model and the first prompt word can be used to generate multiple script segments in segments based on the pre-built video script knowledge base and script framework information, and the initial video script can be generated according to the multiple script segments; the first prompt word includes a script segment prompt word for each script segment in the multiple script segments; finally, the target video script can be generated based on the initial video script using the large language model and the second prompt word; the second prompt word is used to guide the large language model to adjust the script data format of the initial video script. In the method provided by the embodiment of the present application, the script framework information can be obtained according to the script requirement description information and combined with the hot spot information, and the large language model and the pre-built video script knowledge base can be used to generate multiple script segments in segments based on the script framework information, and then the initial video script can be generated, which can improve the efficiency and accuracy of video script generation. Subsequently, the initial video script can also be optimized to obtain the target video script, which can further improve the accuracy of video script generation.

[0058] In an exemplary embodiment, Figure 3 As shown, before step 106, the following steps may be included to provide a method for constructing a video script knowledge base, and the steps specifically include steps 302 to 308. Among them:

[0059] Step 302, obtaining multiple historical video scripts in the historical time period corresponding to the current time period.

[0060] Step 304, obtaining multiple script elements according to the historical video script information of each historical video script in the multiple historical video scripts.

[0061] Step 306: Obtain element attribute information of each script element, and obtain element relationships between the script elements according to the element attribute information.

[0062] Step 308: construct a video script knowledge base based on multiple script elements and element relationships between the script elements.

[0063] Specifically, in the process of obtaining multiple historical video scripts, at least one screening index can be set, such as the number of views, the number of likes, the number of comments, the number of shares, the number of creative novelties and the structural integrity, and the multiple historical video scripts can be obtained from a large number of video scripts using the at least one screening index. Next, each of the multiple historical video scripts can be script disassembled and annotated. According to the historical video script information of each historical video script, each historical video script can be divided into three parts: the introduction part, the main content and the ending part. Each part is disassembled and annotated to obtain multiple script elements of each historical script video, as well as the element relationship between the multiple script elements. For the introduction part, (1) the function and role of the introduction part are deeply analyzed, for example, whether it can quickly attract the audience's attention (by creating suspense, raising thought-provoking questions, showing shocking scenes, etc.), effectively introduce the theme (closely related to the theme and naturally transitioning), and successfully create a specific atmosphere (for example, tense, warm, mysterious, etc.). (2) Extract the key elements of the introduction, namely the script elements, including scene setting (detailed record of the location, time, environmental characteristics of the scene, etc.), character presentation (how the character debuts, how the appearance is shaped, behavioral characteristics, etc.), dialogue style (humorous, serious, poetic, etc., as well as specific language techniques used), and the use of music or sound effects (music type, rhythm, melody, uniqueness of sound effects and their role in creating atmosphere). (3) Classify and annotate the extracted key elements, using labels with clear semantics, such as "attention-grabber", "theme-introduction", "mysterious-atmosphere-builder", etc., for subsequent query and analysis. For the main content, (1) Carefully sort out the plot development context of the main content, clarify the main plot nodes, namely the script elements, and their role in the advancement of the story, such as the turning point of the story (the moment when the plot undergoes a major change, changing the direction of the story), and the division of development stages (dividing the main content into different development stages according to the ups and downs and rhythm of the plot). (2) Analyze the character behaviors and dialogue content in each plot node in detail, and extract the core plot information, including the character's goals, actions, decision-making process in the plot, as well as the key information conveyed by the dialogue, emotional attitudes, and changes in character relationships. (3) Accurately identify the emotional change curve in the main content, mark the emotional climax and trough (determined by analyzing the character's emotional expression, the tension of the plot, the background of music and sound effects, etc.), as well as the corresponding plots and lines, and record the emotional type (for example, joy, sadness, anger, excitement, etc.) and intensity changes.(4) Comprehensively extract the visual elements in the main content, including scene transitions (the method, timing, and purpose of transitions), picture composition characteristics (the use and effects of symmetrical composition, asymmetrical composition, rule of thirds composition, etc.), camera movement methods (the use and effects of push shots, pull shots, pan shots, dolly shots, follow shots, etc.), and color use (main color, color contrast, color symbolic meaning, etc.), and make detailed classification and annotation. For the ending, (1) clarify the purpose of the ending, for example, sublimating the theme (elevating the theme to a higher level and triggering the audience to think more deeply), leaving a deep impression (making the audience unforgettable through unique pictures, touching plots, powerful lines, etc.). (2) Extract the key information of the ending, that is, the script elements, including emotional ending methods (for example, a warm and touching ending, humorous replays, a shocking climax ending, etc.), picture presentation (how the brand logo is displayed, close-up shots of the product, freeze-frame pictures of the scene, etc.). (3) Label the script elements at the end, for example, label the emotional endings as "touching-ending", "humorous-ending", "powerful-climax-ending", etc. according to the emotional type. Next, a video script knowledge base can be constructed based on multiple script elements and the element relationships between the script elements. Specifically, first, each of the multiple script elements of each historical video script can be identified as a node, and multiple nodes of the video script knowledge base can be created, such as the scene node and character node of the introduction part, the plot node and emotional node of the main content, the recommendation node and picture node of the ending part, etc., and edges are defined to represent the node relationship between the nodes, that is, the element relationship; next, the nodes can be labeled according to the element attribute information of the script elements corresponding to each node, such as the name, type and function of the element, and a unique identifier can be assigned to each node; further, the node can be established For example, a "lead to" edge is established between the scene node of the introduction part and the starting plot node of the main content to indicate the transition of the plot; an "involved in" edge is established between the character node and the plot node in which it participates to reflect the character's action trajectory in the story; finally, a graph database is used to store the constructed video script knowledge base, and a complete data management mechanism can be established for the video script knowledge base, including a regular data backup strategy to prevent data loss; an update strategy is formulated to clarify how to update the nodes and edges in the knowledge base when new high-quality scripts are added or the information of existing scripts changes; a convenient data query interface is designed to facilitate the rapid query and use of information in the knowledge base in the subsequent script creation process.

[0064] In the method of this embodiment, a pre-built video script knowledge base can be introduced, and a matching video script of the script requirement description information can be obtained from the video script knowledge base. The large language model is used to generate multiple script segments based on the script framework information and the matching video script, and then an initial video script is generated, which can improve the efficiency and accuracy of video script generation.

[0065] In an exemplary embodiment, Figure 4 As shown, step 106 may include steps 402 to 406. Among them:

[0066] Step 402: Determine the script structure according to the script framework information.

[0067] Among them, the script framework information can be parsed using a large language model to obtain a script structure for generating a video script to be generated. The script structure may include the video architecture and key elements of the video script to be generated, such as video direction, video subject or core concept, target audience, emotional tone and visual elements, etc.

[0068] Step 404, based on the script framework information and the multiple script elements of each historical video script in the multiple historical video scripts contained in the video script knowledge base and the element relationship between the script elements, a matching video script that matches the script framework information is screened out from the multiple historical video scripts.

[0069] Among them, the video script knowledge base includes multiple historical video scripts. Each of the multiple historical video scripts corresponds to multiple nodes and multiple edges, wherein the node is used to represent the constituent elements of each historical video script, and each node corresponds to a script element; the edge is used to represent the relationship between nodes, that is, to represent the relationship between script elements. Each of the multiple historical video scripts includes multiple script elements, and the element relationship between the multiple script elements, and each script element is annotated with corresponding element attribute information. Therefore, the similarity between each historical video script and the video script to be generated can be obtained according to the multiple script elements of each historical video script in the multiple historical video scripts, the element attribute information of each script element, and the element relationship and script framework information between each script element, and then, according to the similarity, the matching video script that matches the video script to be generated can be screened out from the multiple historical video scripts. In a possible implementation, the historical video script whose similarity is greater than the preset similarity threshold can be determined as the matching video script of the video script to be generated.

[0070] Step 406 , using the large language model and the first prompt word, according to the script structure and the matching video script, segmentally generate multiple script segments.

[0071] Among them, the first prompt word of the large language model can be set in advance, and the first prompt word includes a script segment prompt word for each script segment in multiple script segments. For example, the first prompt word can be set to "You are an experienced information flow advertising script copywriting expert, good at extracting valuable data from various sources and integrating it into an easy-to-understand form. You have a deep understanding of video content analysis and can accurately grasp the audience's interests and behavior patterns. You will receive a message: reference video creative content: for reference to the character lines / copywriting / creativity, etc.". Next, the matching video script can be divided into multiple matching script segments, for example, it can be divided into an introduction matching script segment, a main matching script segment and an ending matching script segment. Then, a large language model can be used, under the guidance of the first prompt word, in combination with the script structure and the multiple matching script segments, to generate multiple corresponding script segments by segmented imitation, and generate the initial video script based on the multiple script segments, wherein the introduction matching script segment of the matching video script can be imitated according to the script structure to obtain the introduction script segment of the initial video script; the main matching script segment of the matching video script can be imitated according to the script structure to obtain the main script segment of the initial video script; and the ending matching script segment of the matching video script can be imitated according to the script structure to obtain the ending script segment of the initial video script.

[0072] In the method of this embodiment, a large language model and a pre-built video script knowledge base can be used to generate multiple script segments based on the script framework information, and then generate an initial video script, which can improve the efficiency and accuracy of video script generation.

[0073] In an exemplary embodiment, step 404 may include:

[0074] Based on the script framework information and the multiple script elements of each of the multiple historical video scripts contained in the video script knowledge base and the element relationship between the script elements, the similarity between the script framework information and each of the multiple historical video scripts is obtained; the historical video scripts whose similarity is greater than a preset similarity threshold are determined as matching video scripts.

[0075] Among them, the script framework information can be feature extracted to obtain multiple script feature information of the script framework information, and the multiple script feature information of the script framework information can be multiple script feature information of the video script to be generated; the historical script feature information of the corresponding historical video script can also be obtained according to the multiple script elements of each historical video script, the element attribute information of each script element and the element relationship between each script element; according to the multiple script feature information of the video script to be generated and the historical script feature information of each historical video script, the similarity between the script framework information and each historical video script in the multiple historical video scripts is obtained, and the similarity can be used to indicate the degree of association between each historical video script and the video script to be generated, and then, it can indicate the reference value for generating the video script to be generated. In a possible implementation, using a large language model and a first prompt word, based on a pre-built video script knowledge base and the script framework information, in the process of segmenting and generating multiple script segments, this method can be used to screen the matching script segment of each script segment in the video script knowledge base, and then, according to the script segment feature information of each script segment in the script framework information and the corresponding matching script segment, each script segment is segmented. For example, the multiple script segments may include an introduction script segment, a main body script segment, and an ending script segment; each of the multiple historical video scripts in the video script knowledge base may include a historical introduction script segment, a historical main body script segment, and a historical ending script segment. Thus, according to the script framework information, the introduction feature information of the introduction script segment of the video script to be generated, the main body feature information of the main body script segment, and the ending feature information of the ending script segment can be obtained, and the historical introduction feature information, historical main body feature information, and historical ending feature information of each historical video script can be obtained. Next, based on the similarity between the introduction feature information of the video script to be generated and the historical introduction feature information of the historical introduction script segment of each historical video script in the multiple historical video scripts in the video script knowledge base, the matching historical introduction script segments of the introduction script segment of the video script to be generated can be screened out from the multiple historical introduction script segments contained in the video script knowledge base. Using the same method, the matching historical main script segments of the main script segment of the video script to be generated can also be screened out from the multiple historical main script segments contained in the video script knowledge base, and the matching historical ending script segments of the ending script segment of the video script to be generated can also be screened out from the multiple historical ending script segments contained in the video script knowledge base.

[0076] In the method of this embodiment, a matching video script of the video script to be generated can be obtained in the video script knowledge base, and then, based on the matching video script, the video script to be generated can be obtained by segmented imitation, thereby improving the efficiency and accuracy of video script generation.

[0077] In an exemplary embodiment, Figure 5 As shown, step 102 may include steps 502 to 508. Among them:

[0078] Step 502: Obtain associated script keywords according to the script requirement description information.

[0079] Step 504, obtaining the target associated video from the associated video data source according to the associated script keywords.

[0080] First, it is possible to determine whether the script requirement description information contains video direction information based on the script requirement description information; if the script requirement description information contains the video direction information, it is possible to determine whether the script requirement description information contains a reference video link; if the script requirement description information contains the reference video link, the reference video link is subjected to video parsing to obtain script framework information for generating the video script to be generated; if the script requirement description information does not contain the reference video link, the associated script keywords of the video script to be generated are extracted based on the script requirement description information; next, based on the associated script keywords, multiple candidate associated videos corresponding to the video script to be generated can be obtained from the script requirement description information or the associated video data source of the video script to be generated; further, a target associated video with the highest relevance to the video script to be generated can be screened out from the multiple candidate associated videos, and video parsing is performed on the target associated video to obtain script framework information for generating the video script to be generated; if the script requirement description information does not contain the video direction information, it is determined whether the script requirement description information contains hotspot related information; if the script requirement description information If the information contains hotspot related information, then the associated real-time hotspot corresponding to the script requirement description information is obtained from the associated hotspot data source, and the associated hotspot words are screened out based on the associated real-time hotspots. Next, based on the associated hotspot words, multiple candidate associated videos corresponding to the video script to be generated can be obtained from the script requirement description information or the associated video data source of the video script to be generated. Further, the target associated video with the highest relevance to the video script to be generated can be screened out from the multiple candidate associated videos, and the video is parsed for the target associated video to obtain the script framework information for generating the video script to be generated. If the script requirement description information does not contain hotspot related information, then the corresponding script keywords are determined based on the script requirement description information. Next, based on the script keywords, multiple candidate associated videos corresponding to the video script to be generated can be obtained from the script requirement description information or the associated video data source of the video script to be generated. Further, the target associated video with the highest relevance to the video script to be generated can be screened out from the multiple candidate associated videos, and the video is parsed for the target associated video to obtain the script framework information for generating the video script to be generated.

[0081] Step 506: Acquire multi-dimensional features of the target-related video according to the video information of the target-related video.

[0082] The video information includes at least two of the text information, audio information and video image information of the target associated video; and the multidimensional features include at least emotional expression features.

[0083] Specifically, the target-related video can be subjected to multimodal video analysis. Specifically, first, computer vision technology is used to capture dynamic shots in the video (for example, lens movement modes and speed changes such as pulling, shaking, and moving), character expressions (recognition and analysis of expressions such as joy, anger, sorrow, and happiness), and background elements (scene type, environmental atmosphere, key objects, etc.). Next, the lines in the video can be extracted through automatic speech recognition (ASR) technology, and the lines can be subjected to text analysis with the help of natural language processing (NLP) technology, for example, vocabulary usage (industry terms, popular words, colloquial expressions, etc.), sentence structure (matching of long and short sentences, use of complex sentences, etc.), rhetoric (metaphor, personification, parallelism, etc.), and overall language style (humor, seriousness, lyricism, etc.). Furthermore, we can combine image information, audio information and text information to conduct sentiment analysis, determine the emotional tone conveyed by the video (such as cheerfulness, sadness, tension, warmth, etc.) and the emotional change curve (emotional ups and downs, turning points and trends); identify popular culture memes in the video (such as Internet buzzwords, popular film and television animation elements, social hot topics, etc.) and unique visual elements (such as special color matching, unique picture composition, iconic visual symbols, etc.); summarize the theme content of the video (such as story theme, propaganda theme, core events, etc.), sort out the plot development context (plot ins and outs, key plot nodes, story development rhythm, etc.), and clarify character characteristics (character personality, character relationship, character role in the story, etc.). Then, the product information of the target products involved can be integrated to clarify the core selling points of the target products (e.g., application downloads, e-commerce products, etc.) (unique advantages of the product, user pain points solved, differentiated features from competing products, etc.), target users (user age range, hobbies, etc.), (specific actions that users are expected to take, such as downloading applications, paying attention to brands, etc.), and common video features (commonalities such as style, expression techniques, and element use presented in previous successful videos). Finally, based on the product information of the target product and the results of video analysis, the theme direction of the creative framework can be determined, and the overall structure of the creative framework can be planned, including how to attract the audience's attention at the beginning (such as by raising fitness questions related to the target users, showing fitness results comparison, etc.), how to show the product advantages and user experience in the middle body (such as showing the formulation and implementation process of personalized plans in the scenario of users using applications), how to guide user actions at the end (such as providing preferential incentives for downloading applications, showing user reviews, etc.), and the setting of emotional tone (such as being full of vitality and positivity to inspire users' yearning for a healthy life).

[0084] Step 508: Acquire script framework information according to the multi-dimensional features.

[0085] Among them, through the above method, the multi-dimensional features of the target-related video can be obtained according to the multi-format information of the target-related video, and the multi-format information may include text information, audio information and image information. The multi-dimensional features may include but are not limited to product features, video framework features and emotional expression features. Furthermore, the multi-dimensional feature information of the multi-dimensional features can be input into the large language model as input information. Under the guidance of the pre-set prompt words, the large language model obtains the script framework information according to the multi-dimensional feature information of the multi-dimensional features. The script framework information can be used to describe the video framework of the video script to be generated.

[0086] In the method of this embodiment, associated video scripts of the video script to be generated can be screened out from a variety of video sources. Then, based on the associated video scripts and combined with a large language model, script framework information of the video script to be generated can be generated to facilitate subsequent determination of the script structure of the video script to be generated, which is beneficial to improving the efficiency and accuracy of video script generation.

[0087] In an exemplary embodiment, Figure 6 As shown, step 504 may include steps 602 to 606. Among them:

[0088] Step 602, obtaining a plurality of candidate associated videos from an associated video data source according to the associated script keywords.

[0089] Among them, the associated script keywords can be used to represent the video direction and video theme of the video script to be generated. According to the associated script keywords, multiple candidate associated videos associated with the video direction and video theme of the video script to be generated can be screened out in the associated video data source.

[0090] Step 604, obtaining script requirement characteristics according to the script requirement description information.

[0091] Step 606: Filter out a target associated video from a plurality of candidate associated videos according to the script requirement characteristics.

[0092] Among them, the similarity between each candidate associated video and the video script to be generated can be obtained according to the script requirement characteristics and the script characteristics of the associated video script of each candidate associated video in the multiple candidate associated videos, and the candidate associated video with the similarity greater than a preset similarity threshold is determined as the target associated video of the video script to be generated. In a possible implementation, the candidate associated video with the greatest similarity can be determined as the target associated video of the video script to be generated.

[0093] In an exemplary embodiment, Figure 7 As shown, before step 108, steps 702 to 706 may be included. Among them:

[0094] Step 702, determining the initial data format information of the initial video script.

[0095] Step 704, determining the target data format information of the target video script.

[0096] Step 706: Generate a second prompt word according to the initial data format information and the target data format information.

[0097] The second prompt word is used to guide the large language model to adjust the script data format of the initial video script. The script data format may include a script logic sequence and a data structure, and the script data format of the video script to be generated may be set in advance in the second prompt word. The initial data format information may represent the script logic sequence and data structure of the initial video script, and the data structure of the initial video script may be unstructured data. The target data format information may represent the script logic sequence and data structure of the target video script, and the data structure of the target video script may be structured data.

[0098] In the method of this embodiment, the initial video script can be optimized using a large language model to obtain a target video script, which can further improve the accuracy of video script generation.

[0099] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0100] Based on the same inventive concept, the embodiment of the present application also provides a video script generation device for implementing the video script generation method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more video script generation device embodiments provided below can refer to the limitations of the video script generation method above, and will not be repeated here.

[0101] In one embodiment, Figure 8As shown, a video script generation device is provided, including: an information acquisition module 802, a framework acquisition module 804, a first script generation module 806, and a second script generation module 808, wherein:

[0102] Information acquisition module 802, used to acquire script requirement description information;

[0103] A framework acquisition module 804 is used to acquire script framework information according to the script requirement description information;

[0104] A first script generation module 806 is used to segment a plurality of script segments based on a pre-built video script knowledge base and the script framework information using a large language model and a first prompt word, and generate an initial video script according to the plurality of script segments; the first prompt word includes a script segment prompt word for each of the plurality of script segments;

[0105] The second script generation module 808 is used to generate a target video script based on the initial video script by using the large language model and the second prompt word; the second prompt word is used to guide the large language model to adjust the script data format of the initial video script.

[0106] In one embodiment, the first script generation module 806 is also used to: obtain multiple historical video scripts within the historical time period corresponding to the current time period; obtain multiple script elements based on the historical video script information of each of the multiple historical video scripts; obtain element attribute information of each of the script elements, and obtain element relationship between each of the script elements based on the element attribute information; construct the video script knowledge base based on the multiple script elements and the element relationship between each of the script elements.

[0107] In one embodiment, the first script generation module 806 is further used to: determine the script structure according to the script framework information; screen out a matching video script that matches the script framework information from the multiple historical video scripts according to the script framework information and multiple script elements of each of the multiple historical video scripts contained in the video script knowledge base and the element relationship between the script elements; and generate multiple script segments in segments according to the script structure and the matching video script using the large language model and the first prompt word.

[0108] In one embodiment, the first script generation module 806 is also used to: obtain the similarity between the script framework information and each of the multiple historical video scripts based on the multiple script elements of each of the multiple historical video scripts contained in the script framework information and the video script knowledge base and the element relationship between the script elements; and determine the historical video script whose similarity is greater than a preset similarity threshold as the matching video script.

[0109] In one embodiment, the framework acquisition module 804 is also used to: acquire associated script keywords according to the script requirement description information; acquire a target associated video from an associated video data source according to the associated script keywords; acquire multidimensional features of the target associated video according to video information of the target associated video; the video information includes at least two types of text information, audio information and video image information of the target associated video; the multidimensional features at least include emotional expression features; and acquire the script framework information according to the multidimensional features.

[0110] In one embodiment, the framework acquisition module 804 is also used to: acquire multiple candidate associated videos in the associated video data source according to the associated script keywords; acquire script requirement characteristics according to the script requirement description information; and screen out the target associated video from the multiple candidate associated videos according to the script requirement characteristics.

[0111] In one embodiment, the second script generation module 808 is further used to: determine the initial data format information of the initial video script; determine the target data format information of the target video script; and generate the second prompt word based on the initial data format information and the target data format information.

[0112] Each module in the above video script generation device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0113] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Fig. 9As shown. The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store video script generation related data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a video script generation method is implemented.

[0114] Those skilled in the art will understand that Fig. 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0115] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0116] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0117] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0118] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0119] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0120] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0121] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A video script generation method, characterized in that: The method comprises: Get the script requirement description information; According to the script requirement description information, obtain script framework information; Using the large language model and the first prompt word, based on the pre-built video script knowledge base and the script framework information, multiple script segments are generated in segments, and an initial video script is generated according to the multiple script segments; the first prompt word includes a script segment prompt word for each of the multiple script segments; The target video script is generated based on the initial video script by using the large language model and the second prompt word; the second prompt word is used to guide the large language model to adjust the script data format of the initial video script.

2. The method according to claim 1, characterized in that Before the step of using the large language model and the first prompt word to generate multiple script segments based on a pre-built video script knowledge base and the script framework information, the step further includes: Get multiple historical video scripts in the historical time period corresponding to the current time period; Acquire multiple script elements according to the historical video script information of each of the multiple historical video scripts; Acquire element attribute information of each of the script elements, and acquire element relationships between the script elements according to the element attribute information; The video script knowledge base is constructed based on the multiple script elements and the element relationships between the script elements.

3. The method according to claim 1, characterized in that The method utilizes the large language model and the first prompt word, based on a pre-built video script knowledge base and the script framework information, to generate multiple script segments in segments, including: Determine the script structure according to the script framework information; According to the script framework information and the multiple script elements of each of the multiple historical video scripts contained in the video script knowledge base and the element relationship between the script elements, a matching video script matching the script framework information is screened out from the multiple historical video scripts; The large language model and the first prompt word are used to generate a plurality of script segments in segments according to the script structure and the matching video script.

4. The method according to claim 3, characterized in that The method of screening out a matching video script that matches the script framework information from the multiple historical video scripts according to the script framework information and the multiple script elements of each of the multiple historical video scripts contained in the video script knowledge base and the element relationship between the script elements comprises: According to the script framework information and the multiple script elements of each of the multiple historical video scripts contained in the video script knowledge base and the element relationship between the script elements, the similarity between the script framework information and each of the multiple historical video scripts is obtained; The historical video script whose similarity is greater than a preset similarity threshold is determined as the matching video script.

5. The method according to claim 1, characterized in that: The step of obtaining script framework information according to the script requirement description information includes: According to the script requirement description information, obtain the associated script keywords; According to the associated script keywords, a target associated video is obtained from an associated video data source; According to the video information of the target-associated video, a multidimensional feature of the target-associated video is obtained; the video information includes at least two of text information, audio information and video image information of the target-associated video; and the multidimensional feature includes at least an emotion expression feature; The script framework information is obtained according to the multi-dimensional features.

6. The method according to claim 5, characterized in that The step of obtaining a target associated video from an associated video data source according to the associated script keyword includes: According to the associated script keywords, a plurality of candidate associated videos are obtained from an associated video data source; Acquire script requirement characteristics according to the script requirement description information; According to the script requirement characteristics, the target associated video is screened out from a plurality of the candidate associated videos.

7. The method according to any one of claims 1 to 6, characterized in that: Before generating the target video script based on the initial video script by using the large language model and the second prompt word, the method further includes: Determining initial data format information of the initial video script; Determining target data format information of the target video script; The second prompt word is generated according to the initial data format information and the target data format information.

8. A video script generating device, characterized in that: The device comprises: Information acquisition module, used to obtain script requirement description information; A framework acquisition module, used to acquire script framework information according to the script requirement description information; A first script generation module is used to segment a plurality of script segments based on a pre-built video script knowledge base and the script framework information using a large language model and a first prompt word, and generate an initial video script according to the plurality of script segments; the first prompt word includes a script segment prompt word for each of the plurality of script segments; The second script generation module is used to generate a target video script based on the initial video script by using the large language model and the second prompt word; the second prompt word is used to guide the large language model to adjust the script data format of the initial video script.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

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

Images