AI digital human information processing method and system based on intelligent document generation

Abstract

The invention discloses an AI digital human information processing method and system based on intelligent document generation. The method comprises the steps that live broadcast comment content displayed on a live broadcast interface is acquired; determining first clothing commodity information needing to be displayed according to the live broadcast comment content; obtaining a first oral broadcast copywriting content corresponding to the first clothing commodity information; updating figure appearance data and voice data in the first intelligent document by using the first clothing commodity information and the first oral playing copywriting content to obtain a second intelligent document; and according to the second intelligent document, updating and adjusting the clothing and the oral broadcast content of the current AI anchor. By using the scheme of the invention, the interaction rate and the watching experience of visual audiences are improved, so that the live broadcast effectiveness and activeness are improved.

Description

Technical Field

[0001] The present invention relates to digital human information processing technology, and in particular to an AI digital human information processing method and system based on intelligent document generation. Background Art

[0002] AI digital people refer to virtual characters based on artificial intelligence technology, which can interact with humans through natural language processing, speech recognition, machine learning and other AI technologies (such as the question-and-answer interaction with consumers in the intelligent customer service scenario, which can communicate with consumers to answer questions raised by consumers), and can run on a variety of devices and platforms. It can be seen that these virtual characters not only have a high degree of interaction ability, but also can express emotions and learn in a personalized way, so as to provide more humane services and support. Therefore, AI digital people have been applied to a variety of different scenarios, such as intelligent customer service (including online customer service and telephone customer service), education and training, intelligent guides, virtual idols, live streaming and so on.

[0003] At present, in order to use AI digital humans to achieve live streaming and selling products, it is only necessary to input the specified product introduction and sales copy content, the real anchor's video / image information, appearance information, audio information and other content into the AI ​​digital human's multimodal feature fusion generation model, so that the generated AI digital human can simulate the appearance, voice, expression and behavior of the real anchor, and "speak" the product introduction and sales copy content in the form of oral broadcast to introduce and promote the product. However, in this processing method, the oral broadcast content and the product display content / sequence are all set, so it lacks interaction with the live broadcast audience, and it is also impossible to display the corresponding products according to the needs of the live broadcast audience, which will reduce the efficiency of the live broadcast and the interactivity is also low. Summary of the invention

[0004] The present application aims to solve at least one of the technical problems existing in the prior art. To this end, the present application proposes an AI digital human information processing method and system based on intelligent document generation.

[0005] In a first aspect, an embodiment of the present application provides an AI digital human information processing method based on intelligent document generation, the method comprising:

[0006] S1. Obtain the live broadcast comment content displayed on the live broadcast interface;

[0007] S2. Determine the first clothing product information to be displayed according to the live broadcast comment content;

[0008] S3, obtaining the first oral copy content corresponding to the first clothing product information;

[0009] S4, after updating the character appearance data and voice data in the first smart document using the first clothing product information and the first spoken copy content, a second smart document is obtained;

[0010] S5. Update and adjust the current AI anchor's clothing and spoken content based on the second intelligent document.

[0011] In some embodiments, step S3 specifically includes:

[0012] A301. extracting a plurality of first keywords from the first clothing product information;

[0013] A302, using a plurality of first keywords to generate corresponding first oral copy content;

[0014] Alternatively, a first clothing product information is configured with at least two corresponding oral copy files, and the step S3 specifically includes:

[0015] B301. Randomly select a spoken copy file from at least two corresponding spoken copy files;

[0016] B302. Use the content contained in the selected oral copy file as the first oral copy content.

[0017] In some embodiments, step S3 specifically includes:

[0018] C301. When it is determined that the number of times all the oral copywriting files are selected exceeds the first number threshold, extract a plurality of first keywords from the first clothing product information;

[0019] C302. Generate corresponding first oral copy content using a plurality of first keywords.

[0020] In some embodiments, step S4 specifically includes:

[0021] S401, obtaining a first display behavior action corresponding to first clothing product information;

[0022] S402, after updating the character appearance data, voice data and behavior data in the first smart document using the first clothing product information, the first spoken copy content and the first display behavior action, a second smart document is obtained.

[0023] In some embodiments, the method further comprises the following steps:

[0024] S6. Acquire a first background image corresponding to the first clothing commodity information, wherein a timestamp corresponding to the first clothing commodity information is synchronized with a timestamp of the first background image;

[0025] S7. Use the first background image to replace the live video background.

[0026] In some embodiments, step S2 specifically includes:

[0027] S201, according to a set data statistics time period, determining the number of occurrences of different clothing product labels from the acquired live comment content, wherein the clothing product labels correspond to the clothing product information one by one;

[0028] S202, determining a corresponding first priority selection value according to the number of occurrences of the clothing product label;

[0029] S203, displaying different clothing product numbers on the backend central control interface in order from large to small according to the size of the first priority selected value;

[0030] S204. In response to the first click information input from the backend central control interface, the clothing product number that is clicked and selected is determined, and the clothing product information corresponding to the clothing product number that is clicked and selected is determined as the first clothing product information to be displayed.

[0031] In some embodiments, step S202 specifically includes:

[0032] S2021. Identify the sentiment score corresponding to the clothing product number from the acquired live comment content;

[0033] S2022: Determine a corresponding first priority selection value according to the number of occurrences of the clothing product label and the emotion score.

[0034] In some embodiments, step S2022 specifically includes:

[0035] S20221. Obtain the display time interval corresponding to the clothing product label;

[0036] S20222. Determine a first adjustment coefficient corresponding to the display time interval;

[0037] S20223. Determine a corresponding first priority selection value according to the number of occurrences of the clothing product label, the emotion score, and the first adjustment coefficient.

[0038] In some embodiments, step S2 further specifically includes:

[0039] S205: Obtain the display time interval corresponding to the clothing product number, and display the display time interval as the marking information of the clothing product number on the background central control interface.

[0040] In a second aspect, an embodiment of the present application provides an AI digital human information processing system based on intelligent document generation, the system comprising:

[0041] The first display interface includes the live broadcast interface and the backstage central control interface;

[0042] The background processing system includes at least one processor for executing the steps of implementing the above-mentioned AI digital human information processing based on intelligent document generation.

[0043] This application can achieve at least one of the following technical effects: This application solution identifies the content of user comment data in the live broadcast room to determine the clothing product information to be displayed, and then uses it and the corresponding oral copy content to update and adjust the corresponding type of information content of the smart document to obtain a new smart document. Finally, based on the information of the new smart document, the current AI anchor is updated and rendered, so that the currently displayed clothing and oral content correspond to the clothing product information to be displayed, so as to achieve the display of clothing products while voice reading the corresponding product promotion introduction copy. It can be seen that this solution can adjust product display and introduction according to the real-time needs of active users in the current live broadcast room, which can greatly improve the interaction rate and the intuitive viewing experience of the audience, thereby improving the effectiveness and activity of the live broadcast. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments.

[0045] Figure 1 A schematic diagram of the steps of a first embodiment of an AI digital human information processing method based on intelligent document generation is provided for an embodiment of the present application;

[0046] Figure 2 A schematic flow chart of the steps of a second embodiment of an AI digital human information processing method based on intelligent document generation is provided for an embodiment of the present application;

[0047] Figure 3 A schematic diagram of the structure of a first embodiment of an AI digital human information processing system based on intelligent document generation is provided for an embodiment of the present application;

[0048] Figure 4 A structural schematic diagram of a second embodiment of an AI digital human information processing system based on intelligent document generation is provided for the embodiment of the present application. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described through the implementation method with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application.

[0050] In order to extend the duration of live broadcast with goods, increase brand exposure and save manpower operation costs, more and more live broadcast rooms will use AI digital human technology to realize live broadcast with goods. However, for the existing live broadcast with goods technical solutions based on AI digital human technology, it generally uses pre-set solidified oral text content and product display order as the input information for the AI ​​anchor to generate the rendering model, so that the AI ​​anchor simulates the appearance, voice, expression, and behavior of the real anchor, and reads out the relevant recommendation and introduction content of the goods in the form of oral broadcast. It can be seen that this method lacks interaction with the live audience, and it is impossible to recommend and introduce the corresponding products according to the actual needs of most of the current live audience. This greatly reduces the live broadcast activity, and the interactivity is poor, resulting in low consumer experience, thereby reducing the live broadcast efficiency. To this end, the embodiment of the present application designs an AI digital human information processing solution based on intelligent document generation, which can improve the interactivity with the current live audience, improve the live broadcast room activity, and also improve the consumer viewing experience.

[0051] Reference Figure 1 , an embodiment of the present application provides an AI digital human information processing method based on intelligent document generation, the method comprising the following steps.

[0052] S1. Obtaining live commentary content displayed on the live interface. The live commentary content is mainly input and sent by live viewers and displayed on the live interface. Specifically, the live commentary content may at least include clothing product information that needs to be displayed and introduced by the anchor, the intensity of the need, and the satisfaction with the current display and introduction of clothing products.

[0053] S2. Determine the first clothing product information to be displayed according to the live broadcast comment content.

[0054] Specifically, in view of the large number of live broadcast viewers and their inconsistent demands, it is necessary to identify the demands of the live broadcast viewers, so that the first clothing product information to be displayed can satisfy most of the live broadcast viewers. Therefore, step S2 may specifically include the following steps.

[0055] S201. According to a set data statistics time period, the number of occurrences of different clothing product labels is determined from the acquired live broadcast commentary content, wherein the clothing product labels correspond one-to-one to the clothing product information.

[0056] Specifically, in this embodiment, the clothing product number mainly refers to the serial number of the clothing product link, for example, the clothing product of the first link, the clothing product of the second link, etc. Different clothing product numbers correspond to different clothing product information, wherein the clothing product information may at least include the following contents: clothing types, such as tops (long / short-sleeved T-shirts, shirts, chiffon shirts, etc.), summer clothes (long / short skirts, pants, culottes, etc.); material type; color; style type, such as retro, literary, simple, etc. Therefore, by counting the number of occurrences of clothing product numbers appearing in the live comment content, the viewing needs of most live viewers can be determined, wherein the more clothing product numbers that appear, the more clothing product information that more live viewers want to know. As for the data statistical time period, it can be determined according to the frequency of the live comment content, or it can be set as a fixed value. This can be set according to the actual situation, and no specific limitation is made here.

[0057] In some embodiments, the step of determining the number of occurrences of different clothing product labels from the acquired live comment content may specifically include:

[0058] S2011, using the NLP (natural language processing) model to perform recognition processing on the acquired live comment content, and obtain a corresponding recognition result, which includes the clothing product label;

[0059] S2012, counting the number of occurrences of different clothing product labels. Since NLP technology is used to achieve content recognition, the recognition result has high accuracy, thus further improving the recognition accuracy of clothing product labels.

[0060] S202: Determine a corresponding first priority value according to the number of occurrences of the clothing product label.

[0061] Specifically, the method for determining the first priority selection value can be: directly using the number of occurrences of a clothing product label as the first priority selection value corresponding to the clothing product label (i.e., the first clothing product information); or it can be: adjusting the number of occurrences of a clothing product label as needed, and using the adjusted result as the first priority selection value corresponding to the clothing product label.

[0062] In some embodiments, when the live comment content shows that the live audience has a strong demand for the display of a certain clothing product information, for example, the comment content is "I really want to see the display of the nth linked clothing product, urgent!" and similar sentences, it means that the live audience wants the clothing product more, and the order conversion rate will be higher. Therefore, the priority selection value can be adjusted based on this factor to make the clothing product be displayed as soon as possible, thereby further improving the interactivity and improving the order conversion rate. Therefore, for step S202, it can specifically include the following steps.

[0063] S2021. Identify the sentiment score corresponding to the clothing product number from the obtained live comment content, wherein the higher the sentiment score, the stronger the purchase intention of the live viewer. The sentiment score can also be obtained by using the NLP (natural language processing) model to identify and process the obtained live comment content, which will not be repeated here.

[0064] S2022: Determine a corresponding first priority selection value according to the number of occurrences of the clothing product label and the emotion score.

[0065] Specifically, a second adjustment coefficient k2 corresponding to the emotion score is obtained, wherein the emotion score and the second adjustment coefficient are in positive proportion, i.e., the higher the emotion score, the larger the value of the corresponding second adjustment coefficient; then, a first product result between the number of occurrences M of the clothing product label and the second adjustment coefficient k2 is calculated, and this first product result is used as the first priority selection value corresponding to the clothing product label.

[0066] In some embodiments, in consideration of the needs of merchants, the integrity of product display, and avoiding the problem of low live broadcast satisfaction due to high content repetition, the display time interval is also introduced in the process of determining the priority value. This display time interval refers to whether the display time interval between two adjacent displays of clothing products with the same clothing product label meets the requirements. If so, the priority value of the clothing product with the current clothing product label is not changed or increased. Otherwise, the priority value is reduced. Therefore, the step S2022 may specifically include the following steps.

[0067] S20221. Obtain the display time interval corresponding to the clothing product label. Specifically, the display time interval refers to the time interval between the current moment and the end moment of the last display of the same clothing product. For example, if the priority selection value corresponding to the current i-th clothing product label is the largest after counting the number of times, then the time interval between this moment and the end moment of the last display of the clothing product information corresponding to the i-th clothing product label is calculated, which is the display time interval.

[0068] S20222. Determine a first adjustment coefficient corresponding to the display time interval.

[0069] Specifically, if it is found after statistics that the display time interval corresponding to the current i-th clothing product label is greater than the preset time threshold, then there is no need to adjust the priority selection value, otherwise, the priority selection value needs to be lowered. Therefore, step S20222 may specifically include the following steps:

[0070] When it is determined that the display time interval is greater than or equal to the preset time threshold, the first adjustment sub-value is used as the first adjustment coefficient, and at this time, the first adjustment coefficient is 1;

[0071] When it is determined that the presentation time interval is less than the preset time threshold, the second adjustment sub-value is used as the first adjustment coefficient, wherein the second adjustment sub-value is less than the first adjustment sub-value.

[0072] S20223. Determine a corresponding first priority selection value according to the number of occurrences of the clothing product label, the emotion score, and the first adjustment coefficient.

[0073] Specifically, the first priority value is determined according to the number of occurrences of the clothing product label, the second adjustment coefficient k2, and the first adjustment coefficient k1. Further, this step may specifically include: calculating the second product result between the number of occurrences M of the clothing product label, the second adjustment coefficient k2, and the first adjustment coefficient k1, and using the second product result as the first priority value corresponding to the clothing product label.

[0074] S203. Display different clothing product numbers on the backend central control interface in order from large to small according to the size of the first priority selected value.

[0075] S204. In response to the first click information input from the backend central control interface, the clothing product number that is clicked and selected is determined, and the clothing product information corresponding to the clothing product number that is clicked and selected is determined as the first clothing product information to be displayed.

[0076] Specifically, in order to further ensure and improve the suitability and accuracy of the displayed content, it can be displayed in order from large to small on the background central control interface according to the size of the priority value. This is convenient for the central control personnel to view, and can be combined with the historical experience of the central control personnel and other currently known information (such as current inventory, merchant requirements, etc.) to determine whether to display the clothing product information corresponding to the clothing product number with the largest priority value. If so, the central control personnel can click on the corresponding clothing product number. It can be seen that determining the clothing product information that needs to be displayed in this way can avoid the problem of low accuracy caused by traditional manual counting of the number of times, and under the premise of taking into account other considerations (such as inventory, merchant requirements, etc.), it can accurately meet the needs of most live broadcast viewers, and also bring great convenience to the work of the central control personnel and improve work efficiency.

[0077] In some embodiments, step S2 further specifically includes: S205, obtaining the display time interval corresponding to the clothing product label, and displaying the display time interval as the label information of the clothing product label on the background central control interface. In this way, the central control personnel can further determine whether to select the clothing product with the maximum value to be displayed immediately according to its display time interval, so that its display content and order are more in line with the actual situation.

[0078] S3. Obtain the first spoken copy content corresponding to the first clothing product information. Among them, for the spoken copy content corresponding to the first clothing product information, it can recommend and introduce the product according to the preset spoken content, or it can generate the corresponding spoken copy content based on the keywords of the clothing product information using AI document generation technology. The advantage of the former is that the requirements for system performance are relatively low and the information processing efficiency is high, while the advantage of the latter is that for the same clothing product, the spoken content is different each time, avoiding the reduction of the watchability of the live broadcast content due to repeated and outdated spoken content, thereby reducing its activity and / or the number of viewers. Therefore, for the above-mentioned step S3, it may specifically include the following steps.

[0079] ①. First Implementation Method of Step S3

[0080] A301. extracting a plurality of first keywords from the first clothing product information;

[0081] A302. Generate corresponding first oral copy content using a plurality of first keywords.

[0082] Specifically, the above step S302 can utilize NLG (natural language generation) technology to process the inputted first keywords and generate corresponding first spoken copy content.

[0083] ② Second Implementation Method of Step S3

[0084] B301. Randomly select a spoken copy file from at least two corresponding spoken copy files; wherein a first clothing product information is configured with at least two corresponding spoken copy files;

[0085] B302. Use the content contained in the selected oral copy file as the first oral copy content.

[0086] Specifically, for a clothing product, at least two corresponding oral copy files are pre-configured, that is, for a clothing product, there are at least two sets of oral copy files with different expressions. Then, when it is necessary to oral-read content for a certain clothing product, the oral copy file can be randomly selected to let the AI ​​anchor read out the corresponding content. This improves the viewing quality and fun of the live broadcast while taking into account the processing efficiency.

[0087] In some embodiments, considering that the number of oral text files is limited, if the live broadcast duration of the AI ​​anchor is very long, if it is broadcast non-stop for 24 hours, there will still be problems such as low watchability and low activity due to the high repetitiveness of the oral content. Therefore, for the above step S3, it can also specifically include the following implementation steps, namely the third implementation method.

[0088] C301. When it is determined that the number of times all the oral copywriting files are selected exceeds the first number threshold, extract a plurality of first keywords from the first clothing product information;

[0089] C302. Generate corresponding first oral copy content using a plurality of first keywords.

[0090] Specifically, during the live broadcast, when the set oral copy files have been retrieved and used, and the number of times they have been used exceeds the threshold, then the corresponding oral copy content can be generated using keyword generation, so that the AI ​​anchor can introduce and sell clothing products according to this content. It can be seen that by adopting this method to obtain the oral copy content, it is further achieved to improve the viewing and fun of the live broadcast while taking into account the processing efficiency.

[0091] S4. After updating the character appearance data and voice data in the first smart document using the first clothing product information and the first spoken copy content, a second smart document is obtained.

[0092] Specifically, the smart document contains and stores all kinds of information data required for the subsequent generation of AI digital people, such as character appearance data (such as the character's clothing, appearance, etc.), behavioral actions, voice data (which is the voice data obtained after voice conversion according to the text content read out by the AI ​​anchor), voice features (including the speed, intonation, timbre, etc. configured when the AI ​​anchor speaks out, and these voice features are extracted from the audio files of the real anchor), expression data, etc. Therefore, by adjusting and updating the data stored in the smart document, the AI ​​anchor's clothing, voice-read content, expression, actions, speed, intonation, timbre, etc. can be adjusted. Therefore, by determining the clothing product information to be displayed and the product introduction and sales information that needs to be read out by the AI ​​anchor, the corresponding type data in the smart document can be updated, and the AI ​​anchor can be adjusted.

[0093] In some embodiments, since the embodiments of the present application are mainly for displaying, introducing and selling clothing products, when the AI ​​anchor's clothing is changed to the clothing to be displayed, corresponding display actions can be configured to display the front, back, left and right sides of the clothing in multiple directions. At the same time, corresponding display actions can also be arranged in accordance with the style of the clothing to further highlight the advantages of the clothing, thereby increasing the order conversion rate. Therefore, for step S4, it can specifically include the following steps.

[0094] S401: Acquire a first display behavior action corresponding to first clothing product information.

[0095] Specifically, the first display behavior action may form a mapping relationship with the clothing style information included in the first clothing product information, and then the corresponding first display behavior action may be obtained according to the clothing style included in the first clothing product information.

[0096] Furthermore, a clothing style is configured with at least two different display behavior actions, and the step S401 may specifically include: after obtaining the clothing style information contained in the first clothing product information, randomly selecting a display behavior action from the corresponding at least two different display behavior actions as the first display behavior action.

[0097] Further, the step S401 may specifically include: obtaining corresponding at least two different display behavior actions according to the clothing style information contained in the first clothing product information; then configuring corresponding display timestamps for at least two different display behavior actions, the display timestamp is mainly used to characterize the display duration or display start and / or end time of a display behavior action, so that when the AI ​​anchor performs clothing display, for example, a total of 3 display behavior actions are configured, the behavior action of the AI ​​anchor is first configured as the first display behavior action for display, and when the display time is over, the behavior action of the AI ​​anchor is first configured as the second display behavior action for display, and similarly, when its display time is over, the behavior action of the AI ​​anchor is first configured as the third display behavior action for display, and when these three display behavior actions are executed, the introduction of the current clothing product is also ended, or slightly ended (this can be selected according to the actual situation, without special limitation). Alternatively, by setting the display timestamp, the three display behavior actions can be displayed in sequence according to the set order. This can increase the diversity of the display, thereby further improving viewing interest and enhancing the user viewing experience.

[0098] S402, after updating the character appearance data, voice data and behavior data in the first smart document using the first clothing product information, the first spoken copy content and the first display behavior action, a second smart document is obtained.

[0099] S5. Update and adjust the current AI anchor's clothing and spoken content based on the second intelligent document.

[0100] In some embodiments, reference Figure 2 , the method of the embodiment of the present application further includes the following steps:

[0101] S6. Acquire a first background image corresponding to the first clothing commodity information, wherein a timestamp corresponding to the first clothing commodity information is synchronized with a timestamp of the first background image;

[0102] S7. Use the first background image to replace the live video background.

[0103] Specifically, since different clothing items have different clothing styles, in order to highlight the atmosphere of the clothing, the current live video background can be converted into a first background image that matches the style of the clothing item. This can further improve the watchability, fun and interactivity of the live broadcast, thereby improving the efficiency of the live broadcast.

[0104] Reference Figure 3 The embodiment of the present application also provides an AI digital human information processing system based on intelligent document generation, the system comprising:

[0105] The first display interface includes the live broadcast interface and the backstage central control interface;

[0106] The background processing system includes at least one processor for executing the steps of implementing the above-mentioned AI digital human information processing method based on intelligent document generation.

[0107] Since the background processing system of the present system embodiment includes at least one processor for executing the steps of the above method embodiment, the beneficial effects of the system of the present embodiment are the same as those of the above method embodiment and will not be elaborated in detail here.

[0108] Reference Figure 4 The embodiment of the present application also provides an AI digital human information processing system based on intelligent document generation, the system comprising:

[0109] A first acquisition processing unit, used to acquire live broadcast comment content displayed on the live broadcast interface;

[0110] A first determination processing unit, configured to determine first clothing product information to be displayed according to the live comment content;

[0111] A second acquisition processing unit is used to acquire first oral copy content corresponding to the first clothing product information;

[0112] A first update processing unit, configured to update the appearance data and voice data of the character in the first smart document by using the first clothing product information and the first spoken copy content, to obtain a second smart document;

[0113] The second update processing unit is used to update and adjust the current AI anchor's clothing and oral content according to the second intelligent document.

[0114] In some embodiments, the second acquisition processing unit specifically includes:

[0115] A first extraction processing module, used to extract a plurality of first keywords from the first clothing product information;

[0116] A first generation processing module, used to generate corresponding first oral copy content using a plurality of first keywords;

[0117] Alternatively, a first piece of clothing product information is configured with at least two corresponding oral copywriting files, and the second acquisition processing unit specifically includes:

[0118] A first selection processing module is used to randomly select a spoken copy file from at least two corresponding spoken copy files;

[0119] The first determination processing module is used to use the content included in the selected oral copy file as the first oral copy content.

[0120] In some embodiments, the second acquisition processing unit in step specifically includes:

[0121] A second extraction processing module is used to extract a plurality of first keywords from the first clothing product information when it is determined that the number of times all the oral copywriting files are selected exceeds the first number threshold;

[0122] The second generation processing module is used to generate corresponding first oral copy content using a plurality of first keywords.

[0123] In some embodiments, the first update processing unit specifically includes:

[0124] A first acquisition processing module, used to acquire a first display behavior action corresponding to the first clothing product information;

[0125] The first update processing module is used to update the character appearance data, voice data and behavior data in the first smart document by using the first clothing product information, the first spoken copy content and the first display behavior action to obtain the second smart document.

[0126] In some embodiments, the system further includes:

[0127] A third acquisition processing unit, configured to acquire a first background image corresponding to the first clothing commodity information, wherein a timestamp corresponding to the first clothing commodity information is synchronized with a timestamp of the first background image;

[0128] The third updating processing unit is used to replace the live video background with the first background image.

[0129] In some embodiments, the first determination processing unit specifically includes:

[0130] A first statistical processing module is used to determine the number of occurrences of different clothing product labels from the acquired live comment content according to a set data statistical time period, wherein the clothing product labels correspond to the clothing product information one by one;

[0131] A second determination processing module, used to determine the corresponding first priority selection value according to the number of occurrences of the clothing product label;

[0132] A first display processing module, used to display different clothing product labels in order from large to small according to the size of the first priority selection value on the background central control interface;

[0133] The third determination processing module is used to determine the clothing product label that is clicked and selected in response to the first click information input from the background central control interface, and determine the clothing product information corresponding to the clothing product label that is clicked and selected as the first clothing product information to be displayed.

[0134] In some embodiments, the second determination processing module specifically includes:

[0135] The first recognition processing submodule is used to recognize the emotion score corresponding to the clothing product number from the acquired live comment content;

[0136] The first determination processing submodule is used to determine the corresponding first priority selection value according to the number of occurrences of the clothing product label and the emotion score.

[0137] In some embodiments, the first determination processing submodule specifically includes:

[0138] The first acquisition processing submodule is used to acquire the display time interval corresponding to the clothing product label;

[0139] A second determination processing submodule is used to determine a first adjustment coefficient corresponding to the display time interval;

[0140] The third determination processing submodule is used to determine the corresponding first priority selection value according to the number of occurrences of the clothing product label, the emotion score and the first adjustment coefficient.

[0141] In some embodiments, the first determination processing unit further specifically includes:

[0142] The first display processing unit is used to obtain a display time interval corresponding to the clothing product label, and display the display time interval as marking information of the clothing product label on the background central control interface.

[0143] Since the functional modules of the system embodiment correspond one-to-one with the steps of the above method embodiment, the beneficial effects of the system of the present embodiment are the same as the beneficial effects of the above method embodiment, and will not be elaborated in detail here.

[0144] In addition, the present application also provides a computer-readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the above method embodiment.

[0145] For the processors mentioned in the above storage medium embodiment and the system embodiment, the number can be at least 1, and at least any step in the above method embodiment can be executed. When the number is at least two, at least two processors can be connected to each other by communication, not limited to wired or wireless communication connection, and the at least one processor can be connected to various intelligent terminal devices by communication. In addition, the processor can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0146] Finally, it should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0147] Note that the above are only preferred embodiments of the present application and the technical principles used. Those skilled in the art will understand that the present application is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present application. Therefore, although the present application is described in more detail through the above embodiments, the present application is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present application, and the scope of the present application is determined by the scope of the appended claims.

Claims

1. An AI digital human information processing method based on intelligent document generation, characterized in that: The method includes: S1. Obtain the live broadcast comment content displayed on the live broadcast interface; S2. Determine the first clothing product information to be displayed according to the live broadcast comment content; S3, obtaining the first oral copy content corresponding to the first clothing product information; S4, after updating the character appearance data and voice data in the first smart document using the first clothing product information and the first spoken copy content, a second smart document is obtained; S5. Update and adjust the current AI anchor's clothing and spoken content based on the second intelligent document.

2. The method according to claim 1, characterized in that The step S3 specifically includes: A301. extracting a plurality of first keywords from the first clothing product information; A302, using a plurality of first keywords to generate corresponding first oral copy content; Alternatively, a first clothing product information is configured with at least two corresponding oral copy files, and the step S3 specifically includes: B301. Randomly select a spoken copy file from at least two corresponding spoken copy files; B302. Use the content contained in the selected oral copy file as the first oral copy content.

3. The method according to claim 1, characterized in that The step S3 specifically includes: C301. When it is determined that the number of times all the oral copywriting files are selected exceeds the first number threshold, extract a plurality of first keywords from the first clothing product information; C302. Generate corresponding first oral copy content using a plurality of first keywords.

4. The method according to claim 1, characterized in that The step S4 specifically includes: S401, obtaining a first display behavior action corresponding to first clothing product information; S402: After updating the character appearance data, voice data and behavior data in the first smart document using the first clothing product information, the first spoken copy content and the first display behavior action, a second smart document is obtained.

5. The method according to claim 1, characterized in that The method further comprises the following steps: S6. Acquire a first background image corresponding to the first clothing commodity information, wherein a timestamp corresponding to the first clothing commodity information is synchronized with a timestamp of the first background image; S7. Use the first background image to replace the live video background.

6. The method according to any one of claims 1 to 5, characterized in that: The step S2 specifically includes: S201, according to a set data statistics time period, determining the number of occurrences of different clothing product labels from the acquired live comment content, wherein the clothing product labels correspond to the clothing product information one by one; S202, determining a corresponding first priority selection value according to the number of occurrences of the clothing product label; S203, displaying different clothing product numbers on the backend central control interface in order from large to small according to the size of the first priority selected value; S204. In response to the first click information input from the backend central control interface, the clothing product number that is clicked and selected is determined, and the clothing product information corresponding to the clothing product number that is clicked and selected is determined as the first clothing product information to be displayed.

7. The method according to claim 6, characterized in that The step S202 specifically includes: S2021. Identify the sentiment score corresponding to the clothing product number from the acquired live comment content; S2022: Determine a corresponding first priority selection value according to the number of occurrences of the clothing product label and the emotion score.

8. The method according to claim 7, characterized in that The step S2022 specifically includes: S20221. Obtain the display time interval corresponding to the clothing product label; S20222. Determine a first adjustment coefficient corresponding to the display time interval; S20223. Determine a corresponding first priority selection value according to the number of occurrences of the clothing product label, the emotion score, and the first adjustment coefficient.

9. The method according to claim 6, characterized in that The step S2 further specifically includes: S205: Obtain the display time interval corresponding to the clothing product number, and display the display time interval as the marking information of the clothing product number on the background central control interface.

10. An AI digital human information processing system based on intelligent document generation, characterized in that: The system includes: The first display interface includes the live broadcast interface and the backstage central control interface; A background processing system comprises at least one processor for executing the method according to any one of claims 1 to 9.

Images