AI live broadcast method, device and system

Abstract

The present invention relates to the field of computers, and in particular to an AI live broadcast method, device and system, wherein the method can infer corresponding negative content through a sudden drop in traffic in a traffic curve of a live broadcast of a similar product in history, and then determine whether there is corresponding negative content in the live broadcast copy of this live broadcast. If so, the negative content can be corrected in time, and then the influence of the negative content on the live broadcast traffic can be eliminated, thereby avoiding a sudden drop in the live broadcast traffic as much as possible, ensuring the publicity effect of the product, and then ensuring the sales volume of the product.

Description

Technical Field

[0001] The present invention relates to the field of computers, and in particular to an AI live broadcast method, device and system. Background Art

[0002] The live streaming e-commerce industry has developed rapidly around the world in recent years and has become an important part of the Internet economy. With the development of AI technology, more and more users are generating AI virtual anchors to conduct live streaming e-commerce, which greatly saves human resources.

[0003] The existing AI live broadcast method mainly requires users to input live broadcast copy in advance, and generate live broadcast video with the live broadcast copy (in the live broadcast video, the virtual AI anchor expresses the live broadcast copy in the voice and tone set by the user), and then broadcast the generated live broadcast video through the live broadcast platform to achieve live broadcast with goods; however, during the live broadcast process, the quality of the live broadcast copy set by the user directly affects the live broadcast traffic, and then affects the sales of the product, but users often have strong subjectivity when designing live broadcast copy, and cannot obtain corresponding audience feedback in time. The length of time will cause the good copy that users think will have a negative impact on the audience during the actual live broadcast, which will lead to a sudden decline in live broadcast traffic.

[0004] Therefore, the existing AI live broadcast method has the problem of being unable to timely detect negative content in the live broadcast copy (i.e., content that will cause a sudden decline in traffic) before the live broadcast, and to make timely corrections to the negative content. Summary of the invention

[0005] Based on this, it is necessary to provide an AI live broadcast method, device and system to address the above-mentioned problems.

[0006] The embodiment of the present invention is implemented as follows: an AI live broadcast method, the method comprising:

[0007] S1: Obtain product information and live broadcast copy sent by the user, and determine the product category of the product to be promoted based on the product information;

[0008] S2: retrieve the traffic curves of all live broadcasts of products in the same product category as the product to be promoted within the set time period in the past;

[0009] S3: Perform curve analysis on each flow curve to determine whether there is a sudden flow drop section;

[0010] S4: If there is no sudden drop in traffic, an AI video is generated according to the live broadcast copy, and the AI ​​video is broadcasted through the live broadcast platform;

[0011] S5: If there is a sudden traffic drop segment, for each sudden traffic drop segment, a local video segment corresponding to the sudden traffic drop segment is intercepted in the corresponding live video;

[0012] S6: Determine the negative content in the copy content of each partial video clip, and compare the negative content with the content of the live copy to determine whether there is negative content in the live copy;

[0013] S7: If there is no negative content in the live broadcast copy, generate an AI video based on the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform;

[0014] S8: If there is negative content in the live broadcast copy, determine the replacement content of the negative content, and then replace the negative content in the live broadcast copy with the replacement content to correct the live broadcast copy, generate an AI video based on the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform.

[0015] In one embodiment, the present invention provides an AI live broadcast device, the device comprising:

[0016] The acquisition module is used to obtain the product information and live broadcast copy sent by the user, and determine the product category of the product to be promoted based on the product information;

[0017] A retrieval module is used to retrieve the traffic curves of all live broadcasts of products that are consistent with the product category of the product to be promoted within a set time period in the past;

[0018] The first processing module is used to perform curve analysis on each flow curve to determine whether there is a sudden flow drop section;

[0019] The second processing module is used to generate an AI video according to the live broadcast copy if there is no sudden drop in traffic, and broadcast the AI ​​video through the live broadcast platform;

[0020] The third processing module is used for intercepting a local video segment corresponding to the sudden traffic drop segment in the corresponding live video for each sudden traffic drop segment if there is one;

[0021] The fourth processing module is used to determine the negative content in the copy content of each partial video segment, and compare the negative content with the content of the live copy to determine whether there is negative content in the live copy;

[0022] The fifth processing module is used to generate an AI video based on the live broadcast copy if there is no negative content in the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform;

[0023] The sixth processing module is used to determine the replacement content for the negative content if there is negative content in the live broadcast copy, and then replace the negative content in the live broadcast copy with the replacement content to form a revised live broadcast copy, and generate an AI video based on the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform.

[0024] In one embodiment, the present invention provides an AI live broadcast system, the system comprising:

[0025] The user end is used to send product information and live broadcast copy;

[0026] Storage end, used to store historical live broadcast information;

[0027] The live broadcast end is connected to the user end and the storage end, and is used to execute the AI ​​live broadcast method.

[0028] The present invention provides an AI live broadcast method, device and system, wherein the method comprises obtaining product information and live broadcast copy sent by a user end, and determining the product category of a product to be promoted according to the product information; retrieving the live broadcast flow curves of all products of the same product category as the product to be promoted within a set time period in the past; performing curve analysis on each flow curve to determine whether there is a sudden drop in flow; if there is no sudden drop in flow, generating an AI video according to the live broadcast copy, and broadcasting the AI ​​video through a live broadcast platform; if there is a sudden drop in flow, for each sudden drop in flow, intercepting a local video segment corresponding to the sudden drop in flow in the corresponding live video; determining the negative content in the copy content of each local video segment, and comparing the negative content with the content of the live broadcast copy to determine whether there is negative content in the live broadcast copy; If there is no negative content in the live broadcast copy, an AI video is generated based on the live broadcast copy, and the AI ​​video is broadcast through the live broadcast platform; if there is negative content in the live broadcast copy, the replacement content for the negative content is determined, and then the negative content in the live broadcast copy is replaced with the replacement content to correct the live broadcast copy, and an AI video is generated based on the live broadcast copy, and the AI ​​video is broadcast through the live broadcast platform; in the present application, the corresponding negative content can be inferred through the sudden drop in traffic in the traffic curve of the live broadcast of similar products in history, and then it can be determined whether there is corresponding negative content in the live broadcast copy of this live broadcast. If so, these negative contents can be corrected in time, and then the impact of these negative contents on the live broadcast traffic is eliminated, thereby avoiding a sudden decline in the live broadcast traffic as much as possible, ensuring the publicity effect of the product, and then ensuring the sales of the product. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a first flow chart of an AI live broadcast method provided in one embodiment;

[0030] Figure 2 A second flow chart of an AI live broadcast method provided in one embodiment;

[0031] Figure 3 A schematic diagram of a traffic curve of an AI live broadcast method provided in one embodiment;

[0032] Figure 4A module flow chart of an AI live broadcast device provided in one embodiment;

[0033] Figure 5 A schematic diagram of the composition of an AI live broadcast system provided in one embodiment;

[0034] Figure 6 It is a block diagram of the internal structure of the live broadcast terminal in one embodiment. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention 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 invention and are not intended to limit the present invention.

[0036] It is understood that the terms "first", "second", etc. used in the present invention may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, a first xx script may be referred to as a second xx script, and similarly, a second xx script may be referred to as a first xx script without departing from the scope of the present invention.

[0037] like Figure 1-Figure 2 As shown, in one embodiment, an AI live broadcast method is proposed, the method comprising:

[0038] S1: Obtain product information and live broadcast copy sent by the user, and determine the product category of the product to be promoted based on the product information;

[0039] S2: retrieve the traffic curves of all live broadcasts of products in the same product category as the product to be promoted within the set time period in the past;

[0040] S3: Perform curve analysis on each flow curve to determine whether there is a sudden flow drop section;

[0041] S4: If there is no sudden drop in traffic, an AI video is generated according to the live broadcast copy, and the AI ​​video is broadcasted through the live broadcast platform;

[0042] S5: If there is a sudden traffic drop segment, for each sudden traffic drop segment, a local video segment corresponding to the sudden traffic drop segment is intercepted in the corresponding live video;

[0043] S6: Determine the negative content in the copy content of each partial video clip, and compare the negative content with the content of the live copy to determine whether there is negative content in the live copy;

[0044] S7: If there is no negative content in the live broadcast copy, generate an AI video based on the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform;

[0045] S8: If there is negative content in the live broadcast copy, determine the replacement content of the negative content, and then replace the negative content in the live broadcast copy with the replacement content to correct the live broadcast copy, generate an AI video based on the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform.

[0046] In this embodiment, the method is executed in the live broadcast end, which is a terminal for live broadcast, and can be a smart phone, tablet computer, laptop computer, desktop computer, etc.; the live broadcast end is connected to the storage end, and the storage end is used to store the live broadcast information generated during the live broadcast of the live broadcast end, such as live video, corresponding live broadcast traffic (including traffic curve), live broadcast text and audience comments, etc. The live broadcast end can retrieve the required live broadcast information from the storage end to execute this method. The live broadcast end can be a physical server or terminal, or a server cluster composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud servers, cloud databases, cloud storage and CDN; the user end is a terminal used by users, such as smart phones, tablet computers, laptops, etc. The user end communicates with the live broadcast end, so that the user can send product information and live broadcast text to the live broadcast end through the user end.

[0047] In this embodiment, the product information includes product photos, product names, product categories, product descriptions, etc. The live broadcast end can determine the product category after receiving the product information; the live broadcast copy is a copy designed by the user, which is used to promote the product in the product information (i.e., the product to be promoted); the past set time period can be the past six months, one year, or other time periods. In this embodiment, since the audience groups of the same type of products are roughly the same, their sensitivity to different copy is also roughly the same, that is, the negative content that will have a negative impact on the same audience group is also roughly the same. Therefore, using the traffic curve of the live broadcast of products with the same product category as a reference can accurately infer the negative content in the copy;

[0048] In this embodiment, the traffic drop segment is a local traffic curve where the traffic suddenly drops (the drop rate per unit time reaches a preset value); the AI ​​video described in this embodiment is a video generated by an AI virtual anchor to promote the product to be promoted. An AI virtual anchor generator is pre-installed in the live broadcast end, and the user can select suitable anchor elements from it, such as character image, clothing, accent, etc., to generate a specific virtual anchor. Then the live broadcast end can synthesize the input live broadcast text into corresponding voice through the built-in voice synthesis software, and use the AI ​​engine to make the virtual anchor express the synthesized voice, thereby forming an AI video; the live broadcast platform is a third-party platform, and the user can download the live broadcast platform on the live broadcast end. After the live broadcast end generates the AI ​​video, the AI ​​video can be synchronized to the live broadcast platform for live broadcast.

[0049] In the present application, the corresponding negative content can be inferred through the sudden drop in traffic in the traffic curve of the live broadcast of similar products in history, and then it can be determined whether there is corresponding negative content in the live broadcast copy of this live broadcast. If so, these negative contents can be corrected in time, and then the impact of these negative contents on the live broadcast traffic can be eliminated, thereby avoiding a sudden decline in live broadcast traffic as much as possible, ensuring the publicity effect of the product, and then ensuring the sales of the product.

[0050] like Figure 3 As shown, as a preferred embodiment, the flow curve is a curve on the time-flow coordinate system, the starting time point of the flow curve is the starting time point of the corresponding live broadcast, and the ending time point of the flow curve is the ending time point of the corresponding live broadcast;

[0051] Conduct curve analysis on each flow curve to determine whether there is a sudden drop in flow, including:

[0052] S31: For each flow curve, draw a first straight line perpendicular to the horizontal axis of the time-flow coordinate system through the starting point of the flow curve;

[0053] S32: Draw a second straight line parallel to the first straight line, wherein the direction from the first straight line to the second straight line is consistent with the positive direction of the horizontal axis of the time-flow coordinate system, and the interval width between the first straight line and the second straight line is a set width;

[0054] S33: Identify the maximum flow point and the minimum flow point of the local flow curve between the first straight line and the second straight line;

[0055] S34: Calculate the flow difference between the maximum flow point and the minimum flow point, and determine whether the flow difference is greater than a preset flow difference. If so, the local flow curve is a sudden flow drop section. If not, the local flow section is not a sudden flow drop section.

[0056] S35: The first straight line and the second straight line are simultaneously translated in the positive direction of the horizontal axis of the time-flow coordinate system by a set distance, and steps S33 to S35 are executed until the second straight line is moved to a position passing through the end point of the flow area.

[0057] If multiple sudden flow drop sections are identified on the same flow curve, it also includes:

[0058] For each sudden flow drop segment, determine whether the sudden flow drop segment overlaps with other sudden flow drop segments;

[0059] If so, the traffic sudden drop segment and the overlapping traffic sudden drop segment are merged into a new traffic sudden drop segment.

[0060] In this embodiment, the time difference corresponding to the set width is the set time difference, such as 1 minute, 2 minutes or other time difference, the direction from the first straight line to the second straight line is consistent with the positive direction of the horizontal axis of the time-flow coordinate system, that is, the second straight line is on the right side of the first straight line; the time difference corresponding to the set distance of translation can be 1 / 10 of the set time difference, or other time lengths, the smaller the setting, the higher the accuracy; in this embodiment, it is determined whether there is a sudden drop in traffic, that is, whether there is at least one sudden drop in traffic in at least one live broadcast curve; by gradually translating the first straight line and the second straight line from left to right, all possible local traffic curves can be intercepted, and then each local traffic curve can be judged one by one whether it is a sudden drop in traffic, and then all the sudden drop in traffic can be found; furthermore, on the same traffic curve, if a sudden drop in traffic overlaps with another sudden drop in traffic (including partial overlap), it means that the two sudden drop in traffic are continuous sudden drop segments, and a continuous line segment composed of the two sudden drop in traffic is directly used as a new sudden drop in traffic (that is, the line segment between the first straight line of the sudden drop segment on the left and the second straight line of the sudden drop segment on the right);

[0061] As a preferred embodiment, for each traffic sudden drop segment, a local video segment corresponding to the traffic sudden drop segment is intercepted in the corresponding live video, that is, a video segment of the time interval from the start time point to the end time point of the traffic sudden drop segment is intercepted in the corresponding live video;

[0062] Determine the negative content in the copy of each partial video clip, and compare the negative content with the content of the live copy to determine whether there is negative content in the live copy, including:

[0063] S61: for each partial video segment, determining the text content of the partial video segment;

[0064] S62: Identify all keywords in the copy content;

[0065] S63: Obtain all audience comments in the time interval corresponding to the local video segment;

[0066] S64: Calculate the frequency of each keyword in each audience's comments;

[0067] S65: Screen out keywords whose appearance frequency reaches a preset frequency to obtain a keyword set;

[0068] S66: taking a keyword from the keyword set, and sending the keyword and the audience comments including the keyword to the user end, so that the user can judge whether the keyword has a negative impact on the audience according to the audience comments;

[0069] S67: receiving feedback information sent back by the user end. If the feedback information is positive information, the keyword is negative content. If the feedback information is negative information, the keyword is not negative content.

[0070] S68: taking another keyword from the keyword set, sending the keyword and the audience comments including the keyword to the user terminal, and executing steps S67 to S68 until it is completed to determine whether each keyword in the keyword set is negative content;

[0071] S69: Compare each keyword determined to be negative content with the words in the live broadcast copy one by one, and then determine whether the keyword exists in the live broadcast copy.

[0072] The frequency of each keyword in each audience's comments is calculated using the following formula:

[0073]

[0074] in, is the frequency of occurrence, The total number of comments from viewers. The number of audience comments that include the keyword.

[0075] In this embodiment, for each traffic drop segment, the corresponding live video is first retrieved from the storage end, and a local video segment corresponding to the traffic drop segment is captured in the live video screen. The voice content of the local video segment is recognized by an existing voice recognition algorithm to obtain the text content of the local video segment; then the keywords of the text content are determined by a keyword extraction algorithm (a text keyword extraction algorithm based on word clustering); when negative content appears in the live broadcast, a large number of viewers will exit the live broadcast room, causing a sudden drop in traffic, while some viewers will comment on the negative content, and the proportion of comments on the negative content will be relatively high (therefore a preset frequency is set, such as 50%); since the audience's comments on negative content will have a clear direction, these comments and corresponding keywords are sent to the user end, and the user can easily determine whether it is negative content.

[0076] As a preferred embodiment, determining replacement content for negative content and then replacing the negative content in the live copy with the replacement content includes:

[0077] S81: For each keyword included in the live broadcast copy and determined to be negative content, obtain a synonym of the keyword from a vocabulary database;

[0078] S82: Obtaining the copywriting content corresponding to all traffic sudden drop segments, and identifying the keywords of each copywriting content;

[0079] S83: Select a synonym as a comparison synonym;

[0080] S84: comparing the comparison synonyms with the keywords of the copy content of each partial video clip to determine whether the comparison synonyms appear in at least one copy content, and if not, using the comparison synonyms as replacement content, and then replacing the negative content in the live copy with the replacement content;

[0081] S85: If yes, select another synonym as the comparison synonym, and execute steps S84 to S85 until the replacement content replaces the negative content in the live copy.

[0082] In this embodiment, each word in the vocabulary database is found, and words with similar meanings are grouped together. For each keyword determined to be negative content, the group to which the keyword belongs is found, and its synonyms can be found; the synonyms do not appear in any copy content, which means that the synonyms will not cause a sudden drop in traffic. Using them as replacements can most likely ensure that the live broadcast traffic is not affected.

[0083] As a preferred embodiment, broadcasting AI video via a live broadcast platform includes:

[0084] Divide the AI ​​video into several sub-videos;

[0085] Each sub-video is sent to the live broadcast platform for playback one by one in the order of the corresponding time, wherein the next sub-video is sent to the live broadcast platform before the currently playing sub-video is finished.

[0086] When playing each sub-video on the live broadcast platform, it also includes:

[0087] Generate a real-time flow curve and monitor whether there is a sudden drop in flow in the real-time flow curve;

[0088] If there is a sudden drop in traffic, determine the copywriting content corresponding to the sudden drop in traffic;

[0089] Determine the negative content of the copy, and then determine the replacement content for the negative content;

[0090] For each unplayed sub-video, determine whether the sub-copy content corresponding to the sub-video includes the negative content. If so, replace the negative content in the sub-copy content with the replacement content to obtain the corrected sub-copy content, and generate a new sub-video based on the sub-copy content, and use the new video to replace the unplayed sub-video.

[0091] In this embodiment, the AI ​​video is divided into multiple sub-videos, and the sub-videos are sent to the live broadcast platform in different time periods, so that the live broadcast end can correct the sub-videos that have not been played, thereby improving the flexibility of video adjustment, and sending the next sub-video to the live broadcast platform before the currently playing sub-video is finished, thereby ensuring the connectivity of the sending of each sub-video and the continuity of the live broadcast; furthermore, monitoring whether there is a sudden drop in traffic, determining the text content corresponding to the sudden drop in traffic, determining the negative content of the text content, determining the replacement content of the negative content, and generating a sub-video (consistent with the steps of generating an AI video) are consistent with the relevant steps of the aforementioned embodiment and are not repeated here; in this embodiment, there may still be new negative content after the formal live broadcast. At this time, the solution of this embodiment can be used to correct these contents in a timely manner to minimize the loss.

[0092] like Figure 4 As shown, in one embodiment, an AI live broadcast device is provided, the device comprising:

[0093] The acquisition module is used to obtain the product information and live broadcast copy sent by the user, and determine the product category of the product to be promoted based on the product information;

[0094] A retrieval module is used to retrieve the traffic curves of all live broadcasts of products that are consistent with the product category of the product to be promoted within a set time period in the past;

[0095] The first processing module is used to perform curve analysis on each flow curve to determine whether there is a sudden flow drop section;

[0096] The second processing module is used to generate an AI video according to the live broadcast copy if there is no sudden drop in traffic, and broadcast the AI ​​video through the live broadcast platform;

[0097] The third processing module is used for intercepting a local video segment corresponding to the sudden traffic drop segment in the corresponding live video for each sudden traffic drop segment if there is one;

[0098] The fourth processing module is used to determine the negative content in the copy content of each partial video segment, and compare the negative content with the content of the live copy to determine whether there is negative content in the live copy;

[0099] The fifth processing module is used to generate an AI video based on the live broadcast copy if there is no negative content in the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform;

[0100] The sixth processing module is used to determine the replacement content for the negative content if there is negative content in the live broadcast copy, and then replace the negative content in the live broadcast copy with the replacement content to form a revised live broadcast copy, and generate an AI video based on the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform.

[0101] The process of realizing the functions of each module in the AI ​​live broadcast device provided in the embodiment of the present application can be specifically referred to the aforementioned Figure 1 The description of the illustrated embodiment will not be repeated here.

[0102] like Figure 5 As shown, in one embodiment, an AI live broadcast system is provided, the system comprising:

[0103] The user end is used to send product information and live broadcast copy;

[0104] Storage end, used to store historical live broadcast information;

[0105] The live broadcast end is connected to the user end and the storage end, and is used to execute the AI ​​live broadcast method.

[0106] In the present application, the live broadcast end cooperates with the storage end and the user end to execute the AI ​​live broadcast method, which can infer the corresponding negative content through the sudden drop in traffic in the traffic curve of the live broadcast of similar products in history, and then determine whether there is corresponding negative content in the live broadcast copy of this live broadcast. If so, these negative contents can be corrected in time, thereby eliminating the impact of these negative contents on the live broadcast traffic, thereby avoiding a sudden decline in the live broadcast traffic as much as possible, ensuring the publicity effect of the product, and then ensuring the sales of the product.

[0107] Figure 6 FIG. 2 shows an internal structure diagram of a live broadcast terminal in an embodiment. Figure 6 As shown, the live broadcast terminal includes a processor, a memory, a network interface, an input device and a display screen connected through a system bus. Among them, the memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the live broadcast terminal stores an operating system and may also store a computer program. When the computer program is executed by the processor, the processor can implement the AI ​​live broadcast method provided in the embodiment of the present invention. The internal memory may also store a computer program. When the computer program is executed by the processor, the processor can execute the AI ​​live broadcast method provided in the embodiment of the present invention. The display screen of the live broadcast terminal may be a liquid crystal display screen or an electronic ink display screen, and the input device of the live broadcast terminal may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the shell of the live broadcast terminal, or an external keyboard, touchpad or mouse, etc.

[0108] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a partial structure related to the scheme of the present invention, and does not constitute a limitation on the live broadcast terminal to which the scheme of the present invention is applied. The specific live broadcast terminal may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0109] In one embodiment, the AI ​​live broadcast device provided by the embodiment of the present invention can be implemented in the form of a computer program. The computer program can be Figure 6 The memory of the live broadcast end can store various program modules that constitute the AI ​​live broadcast device, for example, Figure 4 The acquisition module, the retrieval module, the first processing module, the second processing module, the third processing module, the fourth processing module, the fifth processing module and the sixth processing module shown. The computer program composed of each program module enables the processor to execute the steps of the AI ​​live broadcast method of each embodiment of the present invention described in this specification.

[0110] For example, Figure 6 The live broadcast end shown can be Figure 4 The acquisition module in the AI ​​live broadcast device shown executes step S1; the live broadcast end can execute step S2 through the calling module; the live broadcast end can execute step S3 through the first processing module; the live broadcast end can execute step S4 through the second processing module; the live broadcast end can execute step S5 through the third processing module; the live broadcast end can execute step S6 through the fourth processing module; the live broadcast end can execute step S7 through the fifth processing module; the live broadcast end can execute step S8 through the sixth processing module.

[0111] In one embodiment, a live broadcast terminal is provided, the live broadcast terminal comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:

[0112] S1: Obtain product information and live broadcast copy sent by the user, and determine the product category of the product to be promoted based on the product information;

[0113] S2: retrieve the traffic curves of all live broadcasts of products in the same product category as the product to be promoted within the set time period in the past;

[0114] S3: Perform curve analysis on each flow curve to determine whether there is a sudden flow drop section;

[0115] S4: If there is no sudden drop in traffic, an AI video is generated according to the live broadcast copy, and the AI ​​video is broadcasted through the live broadcast platform;

[0116] S5: If there is a sudden traffic drop segment, for each sudden traffic drop segment, a local video segment corresponding to the sudden traffic drop segment is intercepted in the corresponding live video;

[0117] S6: Determine the negative content in the copy content of each partial video clip, and compare the negative content with the content of the live copy to determine whether there is negative content in the live copy;

[0118] S7: If there is no negative content in the live broadcast copy, generate an AI video based on the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform;

[0119] S8: If there is negative content in the live broadcast copy, determine the replacement content of the negative content, and then replace the negative content in the live broadcast copy with the replacement content to correct the live broadcast copy, generate an AI video based on the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform.

[0120] 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 processor performs the following steps:

[0121] S1: Obtain product information and live broadcast copy sent by the user, and determine the product category of the product to be promoted based on the product information;

[0122] S2: retrieve the traffic curves of all live broadcasts of products in the same product category as the product to be promoted within the set time period in the past;

[0123] S3: Perform curve analysis on each flow curve to determine whether there is a sudden flow drop section;

[0124] S4: If there is no sudden drop in traffic, an AI video is generated according to the live broadcast copy, and the AI ​​video is broadcasted through the live broadcast platform;

[0125] S5: If there is a sudden traffic drop segment, for each sudden traffic drop segment, a local video segment corresponding to the sudden traffic drop segment is intercepted in the corresponding live video;

[0126] S6: Determine the negative content in the copy content of each partial video clip, and compare the negative content with the content of the live copy to determine whether there is negative content in the live copy;

[0127] S7: If there is no negative content in the live broadcast copy, generate an AI video based on the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform;

[0128] S8: If there is negative content in the live broadcast copy, determine the replacement content of the negative content, and then replace the negative content in the live broadcast copy with the replacement content to correct the live broadcast copy, generate an AI video based on the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform.

[0129] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. 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 performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0130] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0131] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described 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.

[0132] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. An AI live broadcast method, characterized in that: The method comprises: S1: Obtain product information and live broadcast copy sent by the user, and determine the product category of the product to be promoted based on the product information; S2: retrieve the traffic curves of all live broadcasts of products in the same product category as the product to be promoted within the set time period in the past; S3: Perform curve analysis on each flow curve to determine whether there is a sudden flow drop section; S4: If there is no sudden drop in traffic, an AI video is generated according to the live broadcast copy, and the AI ​​video is broadcasted through the live broadcast platform; S5: If there is a sudden traffic drop segment, for each sudden traffic drop segment, a local video segment corresponding to the sudden traffic drop segment is intercepted in the corresponding live video; S6: Determine the negative content in the copy content of each partial video clip, and compare the negative content with the content of the live copy to determine whether there is negative content in the live copy, wherein the negative content is content that will cause a sudden drop in traffic; S7: If there is no negative content in the live broadcast copy, generate an AI video based on the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform; S8: If there is negative content in the live broadcast copy, determine the replacement content of the negative content, and then replace the negative content in the live broadcast copy with the replacement content to correct the live broadcast copy, generate an AI video based on the corrected live broadcast copy, and broadcast the AI ​​video through the live broadcast platform.

2. The AI ​​live broadcast method according to claim 1, characterized in that: The flow curve is a curve on the time-flow coordinate system. The starting time point of the flow curve is the starting time point of the corresponding live broadcast, and the ending time point of the flow curve is the ending time point of the corresponding live broadcast. Conduct curve analysis on each flow curve to determine whether there is a sudden drop in flow, including: S31: For each flow curve, draw a first straight line perpendicular to the horizontal axis of the time-flow coordinate system through the starting time point of the flow curve; S32: Draw a second straight line parallel to the first straight line, wherein the direction from the first straight line to the second straight line is consistent with the positive direction of the horizontal axis of the time-flow coordinate system, and the interval width between the first straight line and the second straight line is a set width; S33: Identify the maximum flow point and the minimum flow point of the local flow curve between the first straight line and the second straight line; S34: Calculate the flow difference between the maximum flow point and the minimum flow point, and determine whether the flow difference is greater than the preset flow difference and the time point corresponding to the maximum flow point is earlier than the time point corresponding to the minimum flow point. If so, the local flow curve is a sudden drop section, if not, the local flow curve is not a sudden drop section. S35: The first straight line and the second straight line are simultaneously translated in the positive direction of the horizontal axis of the time-flow coordinate system by a set distance, and steps S33 to S35 are executed until the second straight line is moved to a position passing through the end time point of the flow curve.

3. The AI ​​live broadcast method according to claim 2, characterized in that: If multiple sudden flow drop sections are identified on the same flow curve, it also includes: For each sudden flow drop segment, determine whether the sudden flow drop segment overlaps with other sudden flow drop segments; If so, the traffic sudden drop segment and the overlapping traffic sudden drop segment are merged into a new traffic sudden drop segment.

4. The AI ​​live broadcast method according to claim 3, characterized in that: For each traffic sudden drop segment, a partial video segment corresponding to the traffic sudden drop segment is intercepted in the corresponding live video, that is, a video segment of the time interval from the start time point to the end time point of the traffic sudden drop segment is intercepted in the corresponding live video; Determine the negative content in the copy of each partial video clip, and compare the negative content with the content of the live copy to determine whether there is negative content in the live copy, including: S61: for each partial video segment, determining the text content of the partial video segment; S62: Identify all keywords in the copy content; S63: Obtain all audience comments in the time interval corresponding to the local video segment; S64: Calculate the frequency of each keyword in each audience's comments; S65: Screen out keywords whose appearance frequency reaches a preset frequency to obtain a keyword set; S66: taking a keyword from the keyword set, and sending the keyword and the audience comments including the keyword to the user end, so that the user can judge whether the keyword has a negative impact on the audience according to the audience comments; S67: receiving feedback information sent back by the user end. If the feedback information is positive information, the keyword is negative content. If the feedback information is negative information, the keyword is not negative content. S68: taking another keyword from the keyword set, sending the keyword and the audience comments including the keyword to the user terminal, and executing steps S67 to S68 until it is completed to determine whether each keyword in the keyword set is negative content; S69: Compare each keyword determined to be negative content with the words in the live broadcast copy one by one, and then determine whether the keyword exists in the live broadcast copy.

5. The AI ​​live broadcast method according to claim 4, characterized in that: The frequency of each keyword in each audience's comments is calculated using the following formula: in, is the frequency of occurrence, The total number of comments from viewers. The number of audience comments that include the keyword.

6. The AI ​​live broadcast method according to claim 4, characterized in that: Determine the replacement content for negative content, and then use the replacement content to replace the negative content in the live copy, including: S81: For each keyword included in the live broadcast copy and determined to be negative content, obtain a synonym of the keyword from a vocabulary database; S82: Obtaining the copywriting content corresponding to all traffic sudden drop segments, and identifying the keywords of each copywriting content; S83: Select a synonym as a comparison synonym; S84: comparing the comparison synonyms with the keywords of the copy content of each partial video clip to determine whether the comparison synonyms appear in at least one copy content, and if not, using the comparison synonyms as replacement content, and then replacing the negative content in the live copy with the replacement content; S85: If yes, select another synonym as the comparison synonym, and execute steps S84 to S85 until the replacement content replaces the negative content in the live copy.

7. The AI ​​live broadcast method according to claim 1, characterized in that: Live streaming of AI videos via live streaming platforms includes: Divide the AI ​​video into several sub-videos; Each sub-video is sent to the live broadcast platform for playback one by one in the order of the corresponding time, wherein the next sub-video is sent to the live broadcast platform before the currently playing sub-video is finished.

8. The AI ​​live broadcast method according to claim 7, characterized in that: When playing each sub-video on the live broadcast platform, it also includes: Generate a real-time flow curve and monitor whether there is a sudden drop in flow in the real-time flow curve; If there is a sudden drop in traffic, determine the copywriting content corresponding to the sudden drop in traffic; Determine the negative content of the copy, and then determine the replacement content for the negative content; For each unplayed sub-video, determine whether the sub-copy content corresponding to the sub-video includes the negative content. If so, replace the negative content in the sub-copy content with the replacement content to obtain the corrected sub-copy content, and generate a new sub-video based on the sub-copy content, and replace the unplayed sub-video with the new sub-video.

9. An AI live broadcast device, characterized in that: The device comprises: The acquisition module is used to obtain the product information and live broadcast copy sent by the user, and determine the product category of the product to be promoted based on the product information; A retrieval module is used to retrieve the traffic curves of all live broadcasts of products that are consistent with the product category of the product to be promoted within a set time period in the past; The first processing module is used to perform curve analysis on each flow curve to determine whether there is a sudden flow drop section; The second processing module is used to generate an AI video according to the live broadcast copy if there is no sudden drop in traffic, and broadcast the AI ​​video through the live broadcast platform; The third processing module is used for intercepting a local video segment corresponding to the sudden traffic drop segment in the corresponding live video for each sudden traffic drop segment if there is one; The fourth processing module is used to determine the negative content in the copy content of each partial video clip, and compare the negative content with the content of the live copy to determine whether there is negative content in the live copy, wherein the negative content is the content that will cause a sudden drop in traffic; The fifth processing module is used to generate an AI video based on the live broadcast copy if there is no negative content in the live broadcast copy, and broadcast the AI ​​video through the live broadcast platform; The sixth processing module is used to determine the replacement content for the negative content if there is negative content in the live broadcast copy, and then replace the negative content in the live broadcast copy with the replacement content to correct the live broadcast copy, and generate an AI video based on the corrected live broadcast copy, and broadcast the AI ​​video through the live broadcast platform.

10. An AI live broadcast system, characterized in that: The system comprises: The user end is used to send product information and live broadcast copy; Storage end, used to store historical live broadcast information; A live broadcast end, connected to a user end and a storage end, and used to execute the AI ​​live broadcast method as described in any one of claims 1 to 8.

Images