A method, apparatus, storage medium and electronic device for generating sliced videos

Abstract

The present application relates to a method, apparatus, storage medium and electronic device for generating sliced videos, and relates to the technical field of video editing. The method includes: obtaining a live video of a target live account within a preset time and the numerical values of at least one live metric within the preset time, and obtaining the number of slices, duration range, aspect ratio and the emphasized live metric of the sliced video to be generated; splitting the live video into multiple video segments; determining the target metric value corresponding to each video segment based on the numerical values of the respective live metrics, and selecting the segments to be combined from the respective video segments according to the target metric values corresponding to the respective video segments; combining the segments to be combined to obtain the final sliced video, and repeating the step of selecting the segments to be combined from the respective video segments until the number of the final sliced videos reaches the number of slices and the slice duration of the final sliced video is within the duration range. The present application has the effect of improving the efficiency of generating sliced videos.

Description

Technical Field

[0001] This application relates to the technical field of video editing, and in particular to a method, apparatus, storage medium, and electronic device for generating sliced videos. Background Art

[0002] A sliced video refers to multiple small segments obtained by splitting a complete video. These small segments can be processed, transmitted, and played independently. In the live streaming e-commerce industry, a sliced video refers to segments that are exciting, highlight key moments, and have strong attraction in a live streaming e-commerce video, which are then edited and processed to form a new video. Among them, sliced videos play multiple key roles in the live streaming e-commerce industry. They help increase the exposure rate of products, enhance the purchasing intention of viewers, and optimize the effect of live streaming content. They are a very effective marketing tool in the live streaming e-commerce industry, which can help merchants and live streamers better promote products, improve sales performance, and establish closer connections with viewers.

[0003] Currently, the commonly used method for producing sliced videos is as follows: Based on an artificial method, multiple segment videos are intercepted from the complete live video content, and then according to the requirements of the sliced videos to be generated, each segment video is edited and synthesized to obtain the corresponding sliced videos. Finally, according to the requirements of the video aspect ratio of the platform where the sliced videos are to be published, the aspect ratio of the sliced videos is adjusted to obtain the finally publishable sliced videos. In this way, once the number of sliced videos required is large, the speed of producing sliced videos by the artificial method is slow, resulting in low efficiency of generating sliced videos. Summary of the Invention

[0004] To improve the efficiency of generating sliced videos, this application provides a method, apparatus, storage medium, and electronic device for generating sliced videos.

[0005] In the first aspect of this application, a method for generating sliced videos is provided, specifically including:

[0006] Obtain the live video of a target live account within a preset time and the numerical values of at least one live metric within the preset time, and obtain the number of slices, duration range, aspect ratio, and the focused live metric of the sliced video to be generated. The live metric is an index for evaluating the live situation, and the focused live metric is the live metric that is mainly considered when generating the sliced video among all the live metrics;

[0007] Split the live video into multiple video segments, and each video segment corresponds to a segment of live speech within the preset time;

[0008] Based on the metric values of each of the live broadcast metrics, determine the target metric value corresponding to each video segment, and select the segments to be combined from each of the video segments according to the target metric values corresponding to each video segment, where the target metric value is the metric value of the focused live broadcast metric of the corresponding video segment;

[0009] Combine each of the segments to be combined to obtain the final sliced video, and repeat the step of selecting the segments to be combined from each of the video segments until the number of the final sliced videos reaches the slicing number, and the slicing duration of the final sliced video is within the duration range.

[0010] By adopting the above technical solution, the live broadcast video and the corresponding metric values are obtained, thereby providing editing materials for subsequent generation of sliced videos, and at the same time, the generation requirements of the sliced videos such as the slicing number, the duration range, the aspect ratio, and the focused live broadcast metric are obtained. Further, the live broadcast video is segmented according to the live broadcast script, and a plurality of video segments are obtained, so that each video segment is relatively complete. Further, according to the target metric value corresponding to the focused live broadcast metric of each video segment, the segments to be combined that meet the generation requirements are selected from each video segment, and each of the segments to be combined is combined, and finally the final sliced videos with the slicing number meeting the generation requirements are obtained, thereby realizing automatic generation of the required sliced videos according to the generation requirements without manual intervention throughout the process, and improving the efficiency of sliced video generation.

[0011] Optionally, the focused live broadcast metrics include popularity metrics, sales amount metrics, and comment number metrics. The selecting the segments to be combined from each of the video segments according to the target metric values corresponding to each video segment specifically includes:

[0012] Determine the corresponding first rating by taking the ratio of the metric value of the popularity metric of a single video segment to the metric value of the popularity metric of the live broadcast video, determine the corresponding second rating by taking the ratio of the sales amount of the sales amount metric of a single video segment to the total sales amount corresponding to the live broadcast video, and determine the third rating by taking the ratio of the number of comments of the comment number metric of a single video segment to the overall number of comments corresponding to the live broadcast video;

[0013] Determine the fourth rating in the dimension of the user portrait of the followers of the target live broadcast account and the fifth rating in the dimension of the target population for live broadcast account sales;

[0014] Select the segments to be combined from each of the video segments according to the first rating, the second rating, the third rating, the fourth rating, and the fifth rating corresponding to the same video segment.

[0015] By adopting the above technical solutions, the higher the first rating, the better the popularity of the corresponding video clip during the live broadcast; the higher the second rating, the better the sales of the corresponding video clip during the live broadcast; the higher the third rating, the more comments there are during the live broadcast of the corresponding video clip, and the stronger the audience interaction. In addition, the higher the combined result of the fourth rating and the fifth rating, the more likely the corresponding video clip is to attract attention from users and convert them to purchase products. Finally, based on the first rating, the second rating, the third rating, the fourth rating, and the fifth rating, the usability of each video clip is comprehensively evaluated, so that the selected clips to be combined are more valuable.

[0016] Optionally, the determination of the fourth rating of the target live account in the dimension of the user portrait of the live account followers and the fifth rating in the dimension of the target population for live account sales specifically includes:

[0017] Based on the followers of the target live account, determine at least one dimension of the user portrait of the followers and determine the first weight factor for each of the user portraits of the followers;

[0018] Sum up and average the first weight factors to obtain the fourth rating in the dimension of the user portrait of the live account followers;

[0019] Based on the purchasers of the target live account, determine at least one dimension of the population portrait and determine the second weight factor for each of the population portraits;

[0020] Sum up and average the second weight factors to obtain the fifth rating in the dimension of the target population for live account sales.

[0021] By adopting the above technical solutions, by determining the first weight factor of the user portrait of the followers in each dimension, the proportion of the followers of the user portrait in the corresponding dimension among the entire follower population can be better evaluated. Finally, averaging the first weight factors to obtain the fourth rating, so as to accurately evaluate the composition of the followers of the target live account; by determining the second weight factor of the population portrait in each dimension, the proportion of the purchasers of the population portrait in the corresponding dimension among the entire purchaser population can be better evaluated. Finally, averaging the second weight factors to obtain the fifth rating, so as to accurately evaluate the composition of the purchaser population of the target live account.

[0022] Optionally, the selection of the clips to be combined from each of the video clips according to the first rating, the second rating, the third rating, the fourth rating, and the fifth rating corresponding to the same video clip specifically includes:

[0023] Based on the first rating, the second rating, the third rating, the fourth rating, and the fifth rating corresponding to the same video clip, determine the corresponding weighted average;

[0024] Select the video segment with the largest weighted average from each of the video segments and determine it as the middle video segment, and select the previous video segment whose video time period is earlier than the video time period corresponding to the middle video segment and the subsequent video segment whose video time period is later than the video time period corresponding to the middle video segment from each of the video segments;

[0025] Calculate the first difference between the weighted average of the middle video segment minus the weighted average of each of the previous video segments, determine the ratio of the first difference to the weighted average of the corresponding previous video segment as the first month-on-month ratio of the corresponding previous video segment, and determine the previous video segment corresponding to the largest first month-on-month ratio among the first month-on-month ratios as the preposed video segment, and the preposed video segment is the video segment combined before the middle video segment;

[0026] Calculate the second difference between the weighted average of each of the subsequent video segments minus the weighted average of the middle video segment, determine the ratio of the second difference to the weighted average of the middle video segment as the second month-on-month ratio of the corresponding subsequent video segment, and determine the subsequent video segment corresponding to the smallest second month-on-month ratio among the second month-on-month ratios as the postposed video segment, and the postposed video segment is the video segment combined after the middle video segment;

[0027] Determine the preposed video segment, the middle video segment, and the postposed video segment as the segments to be combined.

[0028] By adopting the above technical solution, the larger the first month-on-month ratio, the more the middle video segment grows in terms of the weighted average compared with the corresponding previous video segment. In terms of the weighted average, the middle video segment grows the most compared with the preposed video segment. Taking the preposed video segment as the starting video segment enables the audience to gain a more shocking viewing experience when browsing from the preposed video segment to the middle video segment, and the promotion effect is better; the smaller the second month-on-month ratio, in terms of the weighted average, the less the corresponding subsequent video segment decreases compared with the middle video segment. In terms of the weighted average, the postposed video segment decreases the least compared with the middle video segment. Taking the postposed video segment as the ending video segment can prevent the sudden drop in the attraction of the video segment to the audience when the audience browses from the middle video segment to the postposed video segment and ends watching. Thus, the final sliced video obtained by combining the preposed video segment, the middle video segment, and the postposed video segment subsequently can have better attraction to users.

[0029] Optionally, the method further includes:

[0030] Obtain the portrait of the purchasing population corresponding to the products sold in the live video, and obtain the historical product slices viewed by the historical users in the portrait of the purchasing population. Count the number of historical users whose viewing duration of each historical product slice exceeds a preset duration, where the product corresponding to the historical product slice is the same as the product sold in the live video;

[0031] Select the historical product slice with the first number from each of the historical product slices in descending order of the said quantity, and determine it as the easy-to-like product slice;

[0032] Count the product conversion rates of the historical live accounts that released each easy-to-like product slice. Select the historical live account with the second number from each of the historical live accounts in descending order of the said product conversion rate, and determine it as the easy-to-convert account corresponding to the corresponding easy-to-like product slice;

[0033] Determine the first weight of each easy-to-like product slice and the second weight of the corresponding easy-to-convert account. The first weight is the ratio of the quantity corresponding to each easy-to-like product slice to the sum of the quantities corresponding to all easy-to-like product slices, and the second weight is the ratio of the product conversion rate of a single easy-to-convert account corresponding to the easy-to-like product slice to the sum of the product conversion rates of all corresponding easy-to-convert accounts;

[0034] Based on the first weight and the corresponding second weight, perform music scoring on the final sliced video.

[0035] By adopting the above technical solution, the more the number of historical users whose viewing duration of each historical product slice exceeds the preset duration, the more likely the corresponding historical product slice is to attract the historical users in the portrait of the purchasing population, and then the easy-to-like product slice is determined; the higher the product conversion rate, the more likely the historical live account is to arouse the purchase interest of the users in the portrait of the purchasing population through the corresponding easy-to-like product slice, and then the easy-to-convert account corresponding to the corresponding easy-to-like product slice is determined. Finally, by combining the analysis of the first weight and the corresponding second weight, the reference sliced video is determined from each easy-to-like product slice, and then the reasonable music scoring of the final sliced video is realized, so that the final sliced video can better attract users after being released.

[0036] Optionally, the performing music scoring on the final sliced video based on the first weight and the corresponding second weight specifically includes:

[0037] Calculate the first weight product of the first weight of each easy-to-like product slice and the second weight of each corresponding key live account, where the key live account is an easy-to-convert account whose matching degree with the first type of the product sold in the live video is lower than a preset matching degree threshold;

[0038] When the second - type matching degree between the to - be - released slice account and the promoted product is higher than the matching degree threshold, sum up the products of each of the first weights to obtain the sum of the weighted products of the corresponding Yixi product slices, and select the maximum sum of the weighted products from the sums of the weighted products. The to - be - released slice account is the live - broadcast account that needs to release the final sliced video.

[0039] Determine the background music of the Yixi product slice corresponding to the maximum sum of the weighted products as the first background music, and score the final sliced video based on the first background music.

[0040] By adopting the above - mentioned technical solution, the larger the product of the first weight is, the easier it is for the corresponding key live - broadcast account to attract users in the purchase - population portrait to place orders through the Yixi product slice. If the second - type matching degree is higher than the matching degree threshold, it indicates that the live - broadcast room corresponding to the to - be - released slice account is relatively well - matched with the promoted product. Further, the larger the sum of the weighted products is, the easier it is for users in the purchase - population portrait to place orders in the live - broadcast account with relatively poor matching with the promoted product based on the corresponding Yixi product slice, which shows that it is the Yixi product slice itself that promotes the order - placing more, and then indicates that the corresponding Yixi product slice is more likely to attract users in the purchase - population portrait and complete transactions. Finally, use the background music of the Yixi product slice corresponding to this maximum sum of the weighted products as a reference to score the final sliced video, so that after the subsequent final sliced video is released to the to - be - released slice account, it can better attract users to place orders for the promoted product.

[0041] Optionally, the method further includes:

[0042] When the second - type matching degree between the to - be - released slice account and the promoted product is lower than the matching degree threshold, determine the similar live - broadcast accounts among the key live - broadcast accounts corresponding to each Yixi product slice. The similar live - broadcast account is the key live - broadcast account with the largest third - type matching degree with the to - be - released slice account.

[0043] Calculate the product of the second weight of the second weight of each Yixi product slice and the corresponding similar live - broadcast account.

[0044] Select the maximum second - weighted product from the second - weighted products, determine the background music of the Yixi product slice corresponding to the maximum second - weighted product as the second background music, and score the final sliced video based on the second background music.

[0045] By adopting the above technical solution, if the matching degree between the slice account to be sent and the second type of the product to be sold is lower than the matching degree threshold, it indicates that the matching between the live broadcast room attributes corresponding to the slice account to be released and the product to be sold is poor. Then, the second weight product is determined. The larger the second weight product is, the easier it is for similar live broadcast accounts to attract users in the purchase population portrait of the product to be sold through the corresponding Yixi product slices. Finally, the largest second weight product is selected from each second weight product, and the background music of the Yixi product slice corresponding to this largest second weight product is used as a reference for the background music of the final sliced video, so that it is easier to attract users in the purchase population portrait to place an order for this product to be sold when the final sliced video is released to this slice account to be sent.

[0046] In the second aspect of the present application, a sliced video generation device is provided, which specifically includes:

[0047] An information acquisition module, configured to acquire the live broadcast video of a target live broadcast account within a preset time and the index values of at least one live broadcast index within the preset time, and acquire the number of slices, duration range, aspect ratio, and the focused live broadcast index of the sliced video to be generated. The live broadcast index is an index for evaluating the live broadcast situation, and the focused live broadcast index is the live broadcast index that is mainly considered when generating the sliced video among each live broadcast index;

[0048] A video segmentation module, configured to segment the live broadcast video into multiple video segments, and each video segment corresponds to a section of live broadcast speech within the preset time;

[0049] A segment selection module, configured to determine the target index value corresponding to each video segment based on the index values of each live broadcast index, and select the segments to be combined from each video segment according to the target index values corresponding to each video segment. The target index value is the index value of the focused live broadcast index of the corresponding video segment;

[0050] A slice generation module, configured to combine each segment to be combined to obtain a final sliced video, and repeat the step of selecting the segments to be combined from each video segment until the number of the final sliced videos reaches the number of slices, and the slice duration of the final sliced video is within the duration range.

[0051] By adopting the above technical solution, the information acquisition module acquires the live broadcast video, index values, number of slices, duration range, aspect ratio, and the focused live broadcast index within the preset time. Then, the video segmentation module segments the live broadcast video into multiple video segments. Next, the segment selection module selects the segments to be combined from each video segment. Finally, the slice generation segment combines each segment to be combined to obtain a final sliced video.

[0052] In a third aspect of the present application, a computer-readable storage medium is provided. A computer program is stored in the computer-readable storage medium. When the computer program is loaded and executed by a processor, the method steps described in any one of the first aspect are executed.

[0053] In a fourth aspect of the present application, an electronic device is provided, specifically including:

[0054] A processor, a memory, and a computer program stored in the memory and capable of running on the processor. The processor is used to load and execute the computer program stored in the memory, so that the electronic device executes the method described in any one of the first aspect.

[0055] In summary, the present application includes at least one of the following beneficial technical effects: obtaining a live video and corresponding index values, thereby providing editing materials for subsequent generation of sliced videos, and at the same time obtaining the generation requirements of sliced videos such as the number of slices, duration range, aspect ratio, and focused live broadcast metrics. Further, the live video is segmented according to the live broadcast speech, and a plurality of video segments are obtained, so that each video segment is relatively complete. Further, according to the target index values corresponding to the focused live broadcast metrics of each video segment, the to-be-combined segments that meet the generation requirements are selected from each video segment, and each to-be-combined segment is combined, and finally a final sliced video with the number of slices meeting the generation requirements is obtained, thereby realizing that no manual intervention is required throughout the process, and automatically generating the required sliced videos according to the generation requirements, and improving the efficiency of generating sliced videos. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 is a schematic flowchart of a method for generating a sliced video provided by an embodiment of the present application;

[0057] Figure 2 is a schematic flowchart of another method for generating a sliced video provided by an embodiment of the present application;

[0058] Figure 3 is a schematic structural diagram of a device for generating a sliced video provided by an embodiment of the present application;

[0059] Figure 4 is a schematic structural diagram of another device for generating a sliced video provided by an embodiment of the present application.

[0060] Description of the reference numerals: 11, information acquisition module; 12, video segmentation module; 13, segment selection module; 14, sliced video generation module; 15, first background music module; 16, second background music module. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0061] To enable those skilled in the art to better understand the technical solutions in this specification, the following will clearly and completely describe the technical solutions in the embodiments of this specification in conjunction with the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments.

[0062] In the description of the embodiments of this application, words such as "exemplary", "for example", or "for illustration" are used to indicate examples, illustrations, or explanations. Any embodiment or design solution described as "exemplary", "for example", or "for illustration" in the embodiments of this application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of words such as "exemplary", "for example", or "for illustration" is intended to present the relevant concepts in a specific manner.

[0063] In the description of the embodiments of this application, the term "and / or" merely describes the association relationship of associated objects and indicates that three relationships may exist. For example, A and / or B may represent: A exists alone, B exists alone, and A and B exist simultaneously. Additionally, unless otherwise specified, the meaning of the term "plural" refers to two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. Furthermore, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. The terms "include", "comprise", "have" and their variants all mean "including but not limited to", unless otherwise specifically emphasized in another way.

[0064] See Figure 1 , the embodiments of this application disclose a flowchart of a method for generating sliced videos, which can be implemented depending on a computer program or run on a sliced video generation device based on the von Neumann architecture. This computer program can be integrated into an application or run as an independent tool-type application, and specifically includes:

[0065] S101: Obtain the live video of a target live account within a preset time and the numerical values of at least one live metric within the preset time, and obtain the number of slices, duration range, aspect ratio, and the focused live metric of the sliced video to be generated.

[0066] Specifically, in the embodiments of the present application, the target live account is a live account that needs to slice the video during the live broadcast. Additionally, the target live account can be a live account for selling goods on a short video platform. In other embodiments, it can also be a live account for selling goods on an online shopping platform. The live video is the live video of the target live account within the preset time of the current live broadcast. The preset time can be within the first 10 minutes or within the range from the first 20 minutes to the first 10 minutes. The number of goods sold by the host in the live video is at least one.

[0067] Furthermore, the live broadcast metric is a metric for evaluating the live broadcast situation, and the metric value is the specific value of a single live broadcast metric. The live broadcast metrics include, but are not limited to, popularity metrics, sales amount metrics, comment count metrics, and live broadcast duration during the live broadcast, etc., so as to evaluate the live broadcast situation or effect of the live broadcast room from multi-dimensional live broadcast data such as live broadcast popularity, sales amount, and comment count. The focused live broadcast metric is the live broadcast metric that is mainly considered among various live broadcast metrics when generating sliced videos, that is, the dimension of live broadcast data mainly considered when slicing the live video. Exemplarily, if the focused live broadcast metric is the popularity metric, then when slicing the live video, the video corresponding to the time period with higher live broadcast popularity is mainly sliced. Among them, the sliced video refers to multiple small segments obtained by splitting a complete video, and these small segments can be processed, transmitted, and played independently. In the embodiments of the present application, the sliced video refers to a brand-new video formed by cutting out the wonderful and highly attractive segments from the live video. The number of slices is the number of sliced videos generated based on the live video, and the duration range is the interval range where the duration of the generated sliced videos is located.

[0068] In addition, the execution subject of a sliced video generation method disclosed in the embodiments of the present application is a server. The server is indirectly or directly connected to the terminal wirelessly or by wire. An automatic slicing client is installed in the terminal, and the server is the background server of the client, which can specifically be an independent physical server or a server cluster or distributed system composed of multiple physical servers. The terminal can be a tablet computer or a personal computer.

[0069] In an embodiment of the present application, an implementation scenario is as follows: A person opens a client through a terminal and sends a slice request instruction. The server intercepts the live video within a preset time from the video of the current live broadcast of the target live account, and obtains the index values of at least one live broadcast index within the preset time through the live broadcast data statistically counted by the corresponding short video platform. Exemplarily, the live popularity data, comment data, etc. of the target live account within the preset time can be obtained. In other embodiments, the index values of at least one live broadcast index within the preset time can also be obtained through a preset third-party live broadcast data statistics tool. Further, a feasible way to obtain the number of slices, duration range, aspect ratio, and the focused live broadcast index is as follows: A person with a slice requirement inputs the expected number of generated slices, the duration range of the sliced video, the aspect ratio of the sliced video, and the focused live broadcast index in the client through the terminal, and finally the server obtains the relevant information.

[0070] S102: Split the live video into multiple video segments, and each video segment corresponds to a segment of live speech within a preset time.

[0071] Specifically, after obtaining the live video, it is necessary to split this live video into multiple video segments. In an embodiment of the present application, a feasible splitting method is as follows: Through Natural Language Processing (NLP), the overall live speech of the anchor in this live video is split into live speech of a preset number of segments. Exemplarily, the overall live speech is: Today, I bring you a product of xxx, and this product has xxx features, etc. It can be split into two segments of live speech, namely: Today, I bring you a product of xxx. This product has xxx features. Further, according to the start time and end time of each segment of live speech obtained by splitting in the live video, the corresponding video segment is split from this live video. It should be noted that each video segment corresponds to a segment of live speech within a preset time.

[0072] S103: Based on the index values of each live broadcast index, determine the target index value corresponding to each video segment, and select the segments to be combined from each video segment according to the target index values corresponding to each video segment. The target index value is the index value of the focused live broadcast index of the corresponding video segment.

[0073] Specifically, according to the time period corresponding to each video clip, the index values of the focused live broadcast indexes corresponding to the respective live broadcast indexes of the target live broadcast account within the preset time are screened out, that is, the target index values corresponding to each video clip. Further, according to the target index values corresponding to each video clip, the clips to be combined are selected from all the video clips. One feasible implementation method is as follows: The ratio of the index value of a single video clip in terms of the popularity index to the index value of the live video in terms of the popularity index, that is, the ratio of the average popularity value during the video time period corresponding to the video clip to the average popularity value within the preset time, is determined as the corresponding first rating. The higher the first rating, the better the popularity of the corresponding video clip. Similarly, the ratio of the sales amount of the sales amount index of a single video clip to the total sales amount corresponding to the live video is determined as the corresponding second rating in turn, so as to evaluate the live sales situation in the corresponding video clip. The higher the second rating, the better the live sales situation in the corresponding video clip. The ratio of the number of comments of the comment number index of a single video clip to the overall number of comments corresponding to the live video is determined as the third rating, so as to evaluate the audience interaction during the live broadcast of the corresponding video clip. The higher the third rating, the more comments there are during the live broadcast of the corresponding video clip, and the stronger the audience interaction. It should be noted that the popularity value refers to the number of online audiences during the live broadcast.

[0074] Further, the fourth rating and the fifth rating of a single video clip are determined. A feasible determination method is as follows: According to the attention population of the target live broadcast account, at least one dimension of the user portrait of the attention users is determined, and the first weight factor of each user portrait of the attention users is determined; the sum of the first weight factors is averaged to obtain the fourth rating of the dimension of the user portrait of the attention users of the live broadcast account; According to the purchase population of the target live broadcast account, at least one dimension of the population portrait is determined, and the second weight factor of each population portrait is determined; the sum of the second weight factors is averaged to obtain the fifth rating of the dimension of the target population for bringing goods of the live broadcast account. It should be noted that the attention population is the user group that adds attention to the target live broadcast account.

[0075] Specifically, a preset third-party live broadcast analysis tool is used to analyze the followers of the target live broadcast account to determine the follower user portraits in at least one dimension. Further, according to the number distribution of the follower user portraits in each dimension, the first weight factor of the follower user portraits in each dimension is determined. Exemplarily, the follower user portraits in three dimensions of age from 26 to 30 years old, female gender, and annual income above 150,000 are determined. If the proportion of the number of people aged 26 to 30 years old among the followers of the target live broadcast account is 30%, the proportion of the number of female people is 20%, and the proportion of the number of people with an annual income of 150,000 is 15%. Further, according to the order from high to low of the proportion of the number of people, three weight factor ranges of 0.8 to 1, 0.6 to 0.8, and 0.4 to 0.6 can be divided in turn. Finally, the first weight factor of the follower user portrait in the age dimension is taken from the range of 0.8 to 1 and the value is 0.9. By analogy, the first weight factor of the follower user portrait in the gender dimension can be taken as 0.7, and the first weight factor of the follower user portrait in the annual income dimension can be taken as 0.5. Finally, the average value of each first weight factor is calculated to obtain the fourth rating.

[0076] Further, a third-party live broadcast analysis tool is used to analyze the purchase population of the target live broadcast account within a preset time to determine the population portraits in at least one dimension. Further, according to the number distribution of the population portraits in each dimension, the second weight factor of the population portraits in each dimension is determined. Exemplarily, the population portraits in three dimensions of age from 26 to 30 years old, female gender, and annual income above 150,000 are determined. If among the actual purchase population, the proportion of the number of people aged 26 to 30 years old is 20%, the proportion of the number of female people is 30%, and the proportion of the number of people with an annual income of 150,000 is 15%, then the second weight factor of the population portrait in the age dimension is taken as 0.7 from the range of 0.6 to 0.8, the second weight factor of the population portrait in the gender dimension can be taken as 0.8, and the second weight factor of the population portrait in the annual income dimension can be taken as 0.4. Finally, the average value of each second weight factor is calculated to obtain the fifth rating.

[0077] Further, determine the weighted average of the first rating, second rating, third rating, fourth rating, and fifth rating corresponding to a single video clip, so as to comprehensively evaluate the usability of the single video clip in the entire live video. The higher the weighted average, the higher the usability, and the higher the highlight degree of the corresponding video clip. Then, according to the weighted average of each video clip, select the clips to be combined from all video clips, that is, the video clips that need to be assembled into the sliced video. The specific process is as follows: Select the video clip with the largest weighted average from each video clip and determine it as the middle video clip, that is, the video clip with the highest usability and the most popular among the audience in the live video. Further, select the previous video clip whose video period is earlier than that of the middle video clip from each video clip, and select the subsequent video clip whose video period is later than that of the middle video clip.

[0078] Calculate the first difference between the weighted average of the middle video clip and the weighted average of a single previous video clip, and determine the ratio of the first difference to the weighted average of the corresponding previous video clip as the first month-on-month ratio of the corresponding previous video clip. The larger the first month-on-month ratio, the more the weighted average of the middle video clip increases compared to the corresponding previous video clip, and determine the previous video clip corresponding to the largest first month-on-month ratio among all the first month-on-month ratios as the preposed video clip, that is, the video clip combined before the middle video clip. In terms of the weighted average, the middle video clip increases the most compared to the preposed video clip. Taking the preposed video clip as the starting video clip, when the audience browses from the preposed video clip to the middle video clip, they can obtain a more shocking viewing experience, and the promotion effect is better.

[0079] Further, calculate the second difference between the weighted average of a single subsequent video clip and the weighted average of the middle video clip, and determine the ratio of this second difference to the weighted average of the middle video clip as the second month-on-month ratio of the corresponding subsequent video clip. The smaller the second month-on-month ratio, the less the weighted average of the corresponding subsequent video clip decreases compared to the middle video clip, and determine the subsequent video clip corresponding to the smallest second month-on-month ratio among all the second month-on-month ratios as the postposed video clip, that is, the video clip combined after the middle video clip. In terms of the weighted average, the postposed video clip decreases the least compared to the middle video clip. Taking the postposed video clip as the ending video clip, when the audience browses from the middle video clip to the postposed video clip, it can avoid the sudden drop in the attraction of the video clip to the audience and end the viewing. Finally, determine the preposed video clip, the middle video clip, and the postposed video clip as the clips to be combined, which is convenient for subsequent combination to obtain a sliced video that is more attractive to the audience.

[0080] S104: Combine each segment to be combined to obtain the final sliced video, and repeat the step of selecting segments to be combined from each video segment until the number of final sliced videos reaches the number of slices and the slice duration of the final sliced videos is within the duration range.

[0081] Specifically, combine each segment to be combined in sequence to obtain the final sliced video, and if the slice duration of this final sliced video is within the duration range, it can be regarded as the completion of generating a single sliced video. If the slice duration is not within the duration range, then replace the preceding video segment or the succeeding video segment of the intermediate video segment. Exemplarily, replace the preceding video segment with the preceding video segment with the second-highest first ratio. Further, to meet the requirement of the number of slices, repeat the step of selecting segments to be combined from each video segment, that is, select segments to be combined from each video segment again and combine them. The selected objects do not include the intermediate video segment, and finally obtain the second final sliced video, and so on, until the number of final sliced videos reaches the number of slices.

[0082] See Figure 2 , the flowchart of another method for generating sliced videos disclosed in the embodiments of the present application can be implemented depending on a computer program or run on a sliced video generation device based on the von Neumann architecture. This computer program can be integrated in an application or run as an independent tool application, and specifically includes:

[0083] S201: Obtain the live video of the target live account within a preset time and the index values of at least one live index within the preset time, and obtain the number of slices, duration range, aspect ratio, and the emphasized live index of the sliced video to be generated.

[0084] S202: Split the live video into multiple video segments, and each video segment corresponds to a section of live speech within the preset time.

[0085] S203: Based on the index values of each live index, determine the target index value corresponding to each video segment, and select the segments to be combined from the corresponding video segments according to the target index values corresponding to each video segment. The target index value is the index value of the emphasized live index of the corresponding video segment.

[0086] S204: Combine each segment to be combined to obtain the final sliced video, and repeat the step of selecting segments to be combined from each video segment until the number of final sliced videos reaches the number of slices and the slice duration of the final sliced videos is within the duration range.

[0087] Specifically, refer to steps S101 - S104, which will not be elaborated here.

[0088] S205: Obtain the portrait of the purchasing population corresponding to the products sold during the live video, and obtain the historical product slices viewed by the historical users in the portrait of the purchasing population. Count the number of historical users whose viewing duration of each historical product slice exceeds a preset duration. The product corresponding to the historical product slice is the same as the product sold during the live video.

[0089] S206: Select the historical product slice with the first number from each historical product slice in descending order of the number, and determine it as the easy-to-like product slice.

[0090] S207: Count the product conversion rate of the historical live accounts that released each easy-to-like product slice. Select the historical live account with the second number from each historical live account in descending order of the product conversion rate, and determine it as the easy-conversion account corresponding to the corresponding easy-to-like product slice.

[0091] Specifically, use a preset third-party live broadcast analysis tool to determine the portrait of the purchasing population corresponding to the products sold in this live video. Exemplarily, the portrait of the purchasing population is: aged 26 - 30, female, and with an annual income of about 120,000, etc. Further, through the third-party live broadcast analysis tool, it is possible to determine the historical product slices viewed by the historical users in the portrait of the purchasing population. The products sold in the historical product slices are the same as the products sold in this live video. Then, count the number of historical users whose viewing duration of each historical product slice exceeds the preset duration. The larger the number, the more attractive the corresponding historical product slice is to the historical users in the portrait of the purchasing population. Finally, select the historical product slice with the first number from each historical product slice in descending order of the number, and determine it as the easy-to-like product slice, that is, the product slice video that is easily favored by the users in the portrait of the purchasing population. It should be noted that the third-party live broadcast analysis tool is developed and provided by a third-party institution or company independent of the live broadcast platform, and is a software or service platform for monitoring, collecting, analyzing, and interpreting live broadcast data.

[0092] Further, count the product conversion rate of the historical live accounts that released the easy-to-like product slices, that is, the proportion of users who place purchase orders through the product purchase links in the easy-to-like product slices. The higher the product conversion rate, the easier it is for the historical live account to arouse the purchase interest of the users in the portrait of the purchasing population through the corresponding easy-to-like product slice. Then, select the historical live account with the second number from each historical live account in descending order of the product conversion rate, and determine it as the easy-conversion account corresponding to the corresponding easy-to-like product slice, that is, the live account that is easily converted through the corresponding easy-to-like product slice.

[0093] S208: Determine the first weight of each easy-to-like product slice and the second weight of the corresponding easy-conversion account.

[0094] S209: Score the final sliced video based on the first weight and the corresponding second weight.

[0095] Specifically, determine the first weight of each Yixi product slice and the corresponding second weight of the easy-to-convert account. The first weight is the ratio of the quantity corresponding to each Yixi product slice to the sum of the quantities corresponding to all Yixi product slices. The second weight is the ratio of the product conversion rate of a single easy-to-convert account corresponding to the Yixi product slice to the sum of the product conversion rates of all corresponding easy-to-convert accounts.

[0096] Furthermore, determine the first type matching degree between each Yixi product slice and the various easy-to-convert accounts and the live video's promoted product. A feasible determination method is as follows: Use a third-party live broadcast analysis tool to determine the purchase population portrait and the portrait of the followers of the easy-to-convert account. Then map both the purchase population portrait and the portrait of the followers into a high-dimensional space for vectorization to obtain the corresponding vectors. Then calculate the cosine similarity between the two vectors to obtain the corresponding first type matching degree. The greater the cosine similarity, the more similar the portrait of the followers of the easy-to-convert account and the portrait of the purchase population of the promoted product, and the higher the first type matching degree. Exemplarily, if the promoted product is an infant product and the easy-to-convert account is also a live broadcast account focusing on infants, then the matching degree between the two is relatively high. Further, select the key live broadcast accounts with a first type matching degree lower than the preset matching degree threshold from the various easy-to-convert accounts corresponding to a single Yixi product slice, that is, the easy-to-convert accounts with a poor similarity between the portrait of the followers and the portrait of the purchase population of the promoted product. Similarly, the second type matching degree between the slice account to be released and the promoted product can be determined, where the slice account to be released is the live broadcast account that needs to release the final sliced video.

[0097] Calculate the first weight product of the first weight of each Yixi product slice and the second weight of each corresponding key live account. The larger the first weight product, the easier it is for the corresponding key live account to attract users with the purchase population portrait to place orders through the Yixi product slice. If the second type matching degree is higher than the matching degree threshold, it indicates that the live broadcast room corresponding to the slice account to be released is relatively well-matched with the goods to be sold. Then, sum up the first weight products corresponding to each individual Yixi product slice to obtain the sum of weight products. The larger the sum of weight products, the easier it is for users with the purchase population portrait to place orders in the live account with relatively poor matching with the goods to be sold based on the corresponding Yixi product slice. This shows that it is the Yixi product slice itself that promotes order placement, and further indicates that the corresponding Yixi product slice is more likely to attract users with the purchase population portrait and complete transactions. Finally, select the largest sum of weight products from the sums of weight products, and determine the background music of the Yixi product slice corresponding to this largest sum of weight products as the first background music, and use it as a reference for the background music of the final sliced video or directly set the background music of the final sliced video as the first background music, so that after the final sliced video is released on this slice account to be released, it can effectively attract users to place orders and increase the sales volume of the goods to be sold. In other embodiments, it is also possible to determine the words contained in the Yixi product slice corresponding to this largest sum of weight products, and segment and assemble the video segments containing these words from the live video to obtain the assembled sliced video.

[0098] Furthermore, if the second type matching degree between the slice account to be released and the goods to be sold is lower than the matching degree threshold, it indicates that the live broadcast room attribute corresponding to the slice account to be released has relatively poor matching with the goods to be sold. Then, determine the similar live accounts among the key live accounts corresponding to each individual Yixi product slice, that is, the key live account with the largest third type matching degree with the slice account to be released. It can be understood that the similar live account and the slice account to be released are relatively similar. Then, calculate the second weight product of the first weight of each Yixi product slice and the second weight of the corresponding similar live account. The larger the second weight product, the easier it is for the similar live account to attract users with the purchase population portrait to place orders through the corresponding Yixi product slice. Finally, select the largest second weight product from the second weight products, and determine the background music of the Yixi product slice corresponding to this largest second weight product as the second background music, and set the background music of the final sliced video as this second background music. It should be noted that vectorize the user portraits of the key live account and the slice account to be released, and calculate the cosine similarity between the two vectors, and then determine the third type matching degree between the key live account and the slice account to be released.

[0099] The implementation principle of a sliced video generation method according to an embodiment of the present application is as follows: Obtain a live video and corresponding index values, thereby providing editing materials for subsequent generation of sliced videos. At the same time, obtain the generation requirements of the sliced videos, such as the number of slices, duration range, aspect ratio, and the key live index. Further, divide the live video according to the live speech, obtaining multiple video segments, so that each video segment is relatively complete. Further, according to the target index value corresponding to the key live index of each video segment, select the segments to be combined that meet the generation requirements from each video segment, and combine the segments to be combined, finally obtaining the final sliced video with the number of slices meeting the generation requirements, thereby realizing automatic generation of the required sliced videos without manual intervention throughout the process, and improving the efficiency of sliced video generation.

[0100] The following is an embodiment of the apparatus of the present application, which can be used to execute the method embodiment of the present application. For details not disclosed in the apparatus embodiment of the present application, please refer to the method embodiment of the present application.

[0101] Please refer to Figure 3 , which is a schematic structural diagram of the sliced video generation apparatus provided by the embodiment of the present application. The apparatus applied to the sliced video generation can be implemented as all or part of the apparatus through software, hardware, or a combination of both. The apparatus includes an information acquisition module 11, a video segmentation module 12, a segment selection module 13, and a sliced video generation module 14.

[0102] The information acquisition module 11 is configured to obtain the live video of a target live account within a preset time and the index values of at least one live index within the preset time, and obtain the number of slices, duration range, aspect ratio, and the key live index of the sliced video to be generated. The live index is an index for evaluating the live situation, and the key live index is the live index that is mainly considered when generating the sliced video.

[0103] The video segmentation module 12 is configured to divide the live video into multiple video segments, and each video segment corresponds to a section of live speech within the preset time.

[0104] The segment selection module 13 is configured to determine the target index value corresponding to each video segment based on the index values of each live index, and select the segments to be combined from each video segment according to the target index values corresponding to each video segment. The target index value is the index value of the key live index corresponding to the video segment.

[0105] The sliced video generation module 14 is configured to combine the segments to be combined to obtain the final sliced video, and repeat the step of selecting the segments to be combined from each video segment until the number of the final sliced videos reaches the number of slices, and the sliced duration of the final sliced video is within the duration range.

[0106] Optionally, the segment selection module 13 is specifically configured to:

[0107] Determine the ratio of the metric value of the popularity metric of a single video segment to the metric value of the popularity metric of the live video as the corresponding first rating, determine the ratio of the sales amount of the sales amount metric of a single video segment to the total sales amount corresponding to the live video as the corresponding second rating, and determine the ratio of the number of comments of the comment count metric of a single video segment to the overall number of comments corresponding to the live video as the third rating;

[0108] Determine the fourth rating in the dimension of the user portrait of the followers of the target live account and the fifth rating in the dimension of the target population for live account sales;

[0109] Select the segments to be combined from each video segment according to the first rating, second rating, third rating, fourth rating, and fifth rating corresponding to the same video segment.

[0110] Optionally, the segment selection module 13 is specifically configured to:

[0111] Determine at least one dimension of the user portrait of the followers according to the followers of the target live account, and determine the first weight factor for each user portrait;

[0112] Sum and average each first weight factor to obtain the fourth rating in the dimension of the user portrait of the followers of the live account;

[0113] Determine at least one dimension of the population portrait according to the purchase population of the target live account, and determine the second weight factor for each population portrait;

[0114] Sum and average each second weight factor to obtain the fifth rating in the dimension of the target population for live account sales.

[0115] Optionally, the segment selection module 13 is specifically configured to:

[0116] Determine the corresponding weighted average according to the first rating, second rating, third rating, fourth rating, and fifth rating corresponding to the same video segment;

[0117] Select the video segment with the largest weighted average from each video segment and determine it as the intermediate video segment, and select the previous video segment whose video time period is earlier than the video time period corresponding to the intermediate video segment and the subsequent video segment whose video time period is later than the video time period corresponding to the intermediate video segment from each video segment;

[0118] Calculate the first difference by subtracting the weighted average of each previous video segment from the weighted average of the intermediate video segment. Determine the ratio of the first difference to the weighted average of the corresponding previous video segment as the first month-on-month ratio of the corresponding previous video segment, and determine the previous video segment corresponding to the largest first month-on-month ratio among all the first month-on-month ratios as the preposed video segment. The preposed video segment is the video segment combined before the intermediate video segment;

[0119] Calculate the second difference by subtracting the weighted average of the intermediate video segment from the weighted average of each subsequent video segment. Determine the ratio of the second difference to the weighted average of the intermediate video segment as the second month-on-month ratio of the corresponding subsequent video segment, and determine the subsequent video segment corresponding to the smallest second month-on-month ratio among all the second month-on-month ratios as the postposed video segment. The postposed video segment is the video segment combined after the intermediate video segment;

[0120] Determine the preposed video segment, the intermediate video segment, and the postposed video segment as the segments to be combined.

[0121] Optionally, as Figure 4 shown, the device further includes a first music matching module 15, which is specifically configured to:

[0122] Obtain the portrait of the purchasing population corresponding to the product being sold in the live video, and obtain the historical product slices viewed by the historical users in the portrait of the purchasing population. Count the number of historical users whose viewing duration of each historical product slice exceeds a preset duration. The product corresponding to the historical product slice is the same as the product being sold;

[0123] Select the historical product slice with the first number among all the historical product slices in descending order of the number as the popular product slice;

[0124] Count the product conversion rates of the historical live accounts that released each popular product slice, and select the historical live account with the second number among all the historical live accounts in descending order of the product conversion rate as the easy-conversion account corresponding to the corresponding popular product slice;

[0125] Determine the first weight of each popular product slice and the second weight of the corresponding easy-conversion account. The first weight is the ratio of the quantity corresponding to each popular product slice to the sum of the quantities corresponding to all the popular product slices, and the second weight is the ratio of the product conversion rate of a single easy-conversion account corresponding to the popular product slice to the sum of the product conversion rates of all the corresponding easy-conversion accounts;

[0126] Match music for the final sliced video based on the first weight and the corresponding second weight.

[0127] Optionally, the first music matching module 15 is specifically configured to:

[0128] Calculate the first weight product of the first weight of each Yixi product slice and the second weight of each corresponding key live account, where the key live account is an easy-to-convert account with a matching degree lower than the preset matching degree threshold for the first type of the product to be sold;

[0129] When the matching degree of the slice account to be sent and the second type of the product to be sold is higher than the matching degree threshold, sum up the first weight products to obtain the sum of the weight products of the corresponding Yixi product slice, and select the largest sum of the weight products from the sums of the weight products. The slice account to be sent is the live account that needs to publish the final slice video;

[0130] Determine the background music of the Yixi product slice corresponding to the largest sum of the weight products as the first background music, and perform music scoring on the final slice video based on the first background music.

[0131] Optionally, the device further includes a second music scoring module 16, specifically for:

[0132] When the matching degree of the slice account to be sent and the second type of the product to be sold is lower than the matching degree threshold, determine the similar live account among the key live accounts corresponding to each Yixi product slice. The similar live account is the key live account with the largest matching degree for the third type with the slice account to be sent;

[0133] Calculate the second weight product of the first weight of each Yixi product slice and the second weight of the corresponding similar live account;

[0134] Select the largest second weight product from the second weight products, determine the background music of the Yixi product slice corresponding to the largest second weight product as the second background music, and perform music scoring on the final slice video based on the second background music.

[0135] It should be noted that when the slice video generation device provided in the above embodiment executes the slice video generation method, only the above division of each functional module is used for illustration. In practical applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the slice video generation device provided in the above embodiment and the slice video generation method embodiment belong to the same concept, and the implementation process is detailed in the method embodiment, which will not be repeated here.

[0136] The embodiment of the present application also discloses a computer-readable storage medium, and the computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the slice video generation method of the above embodiment is adopted.

[0137] Among them, the computer program can be stored in a computer-readable medium. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some middleware form, etc. The computer-readable medium includes any entity or device, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the computer-readable medium includes, but is not limited to, the above components.

[0138] Among them, through this computer-readable storage medium, the method for generating a sliced video in the above embodiment is stored in the computer-readable storage medium and is loaded and executed on the processor to facilitate the storage and application of the above method.

[0139] The embodiment of the present application also discloses an electronic device. When a computer program stored in the computer-readable storage medium is loaded and executed by the processor, the above method for generating a sliced video is adopted.

[0140] Among them, the electronic device can be a desktop computer, a laptop computer, or a cloud server, etc. And the electronic device includes, but is not limited to, a processor and a memory. For example, the electronic device may also include input / output devices, network access devices, and a bus, etc.

[0141] Among them, the processor can adopt a central processing unit (CPU). Of course, according to the actual usage situation, other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. can also be adopted. The general-purpose processor can adopt a microprocessor or any conventional processor, etc. The present application does not make any restrictions on this.

[0142] Among them, the memory can be an internal storage unit of the electronic device. For example, the hard disk or memory of the electronic device, or it can also be an external storage device of the electronic device. For example, a plug-in hard disk, a smart media card (SMC), a secure digital card (SD), or a flash card (FC), etc. equipped on the electronic device. And the memory can also be a combination of the internal storage unit and the external storage device of the electronic device. The memory is used to store the computer program and other programs and data required by the electronic device. The memory can also be used to temporarily store the data that has been output or will be output. The present application does not make any restrictions on this.

[0143] Among them, through this electronic device, a method for generating a sliced video in the above embodiment is stored in the memory of the electronic device and is loaded and executed on the processor of the electronic device for convenient use.

[0144] The above are only exemplary embodiments of the present disclosure, and the scope of the present disclosure cannot be limited thereby. That is, any equivalent changes and modifications made according to the teachings of the present disclosure still fall within the scope covered by the present disclosure. This application aims to cover any variations, uses, or adaptive changes of the present disclosure, which follow the general principles of the present disclosure and include common general knowledge or conventional technical means in the technical field not recorded in the present disclosure. The description and embodiments are only regarded as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A method for generating sliced ​​video, characterized in that: The method comprises: Obtain the live video of the target live account within a preset time and the indicator value of at least one live indicator within the preset time, and obtain the number of slices, duration range, aspect ratio, and live broadcast indicator of the sliced ​​video to be generated, wherein the live broadcast indicator is an indicator for evaluating the live broadcast situation, and the live broadcast indicator is a live broadcast indicator that is focused on when generating the sliced ​​video among the live broadcast indicators; Dividing the live video into multiple video segments, each of which corresponds to a segment of live speech within the preset time; Based on the indicator values ​​of each of the live broadcast indicators, a target indicator value corresponding to each of the video clips is determined, and according to the target indicator values ​​corresponding to each of the video clips, a segment to be combined is selected from each of the video clips, wherein the target indicator value is an indicator value of the live broadcast indicator focused on the corresponding video clip; The segments to be combined are combined to obtain a final sliced ​​video, and the step of selecting the segments to be combined from the video segments is repeated until the number of the final sliced ​​videos reaches the number of slices, and the slice duration of the final sliced ​​video is within the duration range; wherein the live broadcast indicator includes a popularity indicator, a sales indicator, and a comment number indicator, and the segment to be combined from the video segments according to the target indicator value corresponding to each of the video segments includes: Determine the first rating as the ratio of the index value of the popularity index of the single video clip to the index value of the popularity index of the live video, determine the second rating as the ratio of the sales of the sales index of the single video clip to the total sales of the live video, and determine the third rating as the ratio of the number of comments of the number of comments index of the single video clip to the overall number of comments of the live video; Determine the fourth rating of the target live broadcast account in the dimension of the portrait of the users followed by the live broadcast account and the fifth rating in the dimension of the target group of the live broadcast account; Selecting a segment to be combined from each of the video segments according to the first rating, the second rating, the third rating, the fourth rating, and the fifth rating corresponding to the same video segment, including: determining a corresponding weighted average according to the first rating, the second rating, the third rating, the fourth rating, and the fifth rating corresponding to the same video segment; Select a video segment with the largest weighted average from each of the video segments and determine it as the middle video segment, and select a preceding video segment whose video period is earlier than the video period corresponding to the middle video segment and a succeeding video segment whose video period is later than the video period corresponding to the middle video segment from each of the video segments; Calculating a first difference between the weighted average of the middle video segment and the weighted average of each of the previous video segments, determining a ratio of the first difference to the weighted average of the corresponding previous video segment as a first ring ratio value of the corresponding previous video segment, and determining a previous video segment corresponding to a maximum first ring ratio value among the first ring ratio values ​​as a front video segment, wherein the front video segment is a video segment combined before the middle video segment; Calculating a second difference value obtained by subtracting a weighted average value of the middle video segments from a weighted average value of each of the rear video segments, determining a ratio of the second difference value to the weighted average value of the middle video segments as a second ring ratio value of the corresponding rear video segment, and determining a rear video segment corresponding to a minimum second ring ratio value among the second ring ratio values ​​as a post-positioned video segment, wherein the post-positioned video segment is a video segment combined after the middle video segment; The front video segment, the middle video segment and the rear video segment are determined as segments to be combined.

2. The method for generating sliced ​​video according to claim 1, characterized in that: Determining the fourth rating of the target live broadcast account in the dimension of the portrait of the users followed by the live broadcast account and the fifth rating in the dimension of the target group of the live broadcast account specifically includes: Determine, according to the follower group of the target live broadcast account, a follower user portrait of at least one dimension, and determine a first weight factor for each follower user portrait; The first weight factors are summed and averaged to obtain a fourth rating of the live broadcast account's follower portrait dimension; Determine a crowd portrait of at least one dimension according to the purchasing crowd of the target live broadcast account, and determine a second weight factor for each of the crowd portraits; The sum and average of the second weight factors are calculated to obtain the fifth rating of the target population of the live broadcast account.

3. The method for generating sliced ​​video according to claim 1, characterized in that: The method further comprises: Obtain a purchaser profile corresponding to the product carried in the live video, and obtain historical product slices watched by historical users of the purchaser profile, and count the number of historical users whose viewing time of each historical product slice exceeds a preset time, and the product corresponding to the historical product slice is the same as the product carried; According to the order of the number from largest to smallest, the first number of historical product slices is selected from each of the historical product slices to be determined as the Yixi product slice; Counting the product conversion rate of the historical live broadcast account that publishes each Yixi product slice, and selecting the second historical live broadcast account from each of the historical live broadcast accounts in descending order of the product conversion rate to determine it as the easy conversion account of the corresponding Yixi product slice; Determine a first weight of each Yixi product slice and a second weight of the corresponding Yibianhua account, wherein the first weight is the ratio of the quantity corresponding to each Yixi product slice to the sum of the quantities corresponding to all Yixi product slices, and the second weight is the ratio of the product conversion rate of a single Yibianhua account corresponding to the Yixi product slice to the sum of the product conversion rates of all Yibianhua accounts; Based on the first weight and the corresponding second weight, music is composed for the final sliced ​​video.

4. The method for generating sliced ​​video according to claim 3, characterized in that: The performing music matching for the final sliced ​​video based on the first weight and the corresponding second weight specifically includes: Calculate the first weight product of each of the Yixi product slices and the second weight of each corresponding key live broadcast account, where the key live broadcast account is an easy-to-convert account whose matching degree with the first type of the product is lower than a preset matching degree threshold; When the matching degree between the to-be-sent slice account and the second type of the product is higher than the matching degree threshold, the first weight products are summed to obtain the sum of the weight products of the corresponding Yixi product slices, and the maximum sum of the weight products is selected from the sums of the weight products. The to-be-sent slice account is the live broadcast account that needs to publish the final slice video; The background music of the Yixi product slice corresponding to the sum of the maximum weight products is determined as the first background music, and the final slice video is composed of music based on the first background music.

5. The method for generating sliced ​​video according to claim 4, characterized in that: The method further comprises: When the matching degree between the to-be-sent slice account and the second type of the product is lower than the matching degree threshold, determine a similar live broadcast account among the key live broadcast accounts corresponding to each Yixi product slice, wherein the similar live broadcast account is the key live broadcast account with the greatest matching degree of the third type with the to-be-sent slice account; Calculate the second weight product of the first weight of each Yixi product slice and the second weight of the corresponding similar live broadcast account; The maximum second weight product is selected from each of the second weight products, the background music of the Yixi product slice corresponding to the maximum second weight product is determined as the second background music, and the final slice video is composed of music based on the second background music.

6. A slice video generation device, used to implement the method according to any one of claims 1 to 5, characterized in that: include: An information acquisition module (11) is used to acquire the live video of the target live account within a preset time and the index value of at least one live index within the preset time, and to acquire the number of slices, the duration range, the aspect ratio and the emphasis live index of the sliced ​​video to be generated, wherein the live index is an index for evaluating the live broadcast situation, and the emphasis live index is a live index that is emphasized when generating the sliced ​​video among the live indicators; A video segmentation module (12) is used to segment the live video into a plurality of video segments, each of which corresponds to a segment of live speech within the preset time; A segment selection module (13) is used to determine the target indicator value corresponding to each of the video segments based on the indicator value of each of the live broadcast indicators, and select the segment to be combined from each of the video segments according to the target indicator value corresponding to each of the video segments, wherein the target indicator value is the indicator value of the live broadcast indicator of the corresponding video segment; The slice generation module (14) is used to combine the segments to be combined to obtain a final slice video, and repeatedly perform the step of selecting the segments to be combined from the video segments until the number of the final slice video reaches the number of slices, and the slice duration of the final slice video is within the duration range.

7. A computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, characterized in that: When the computer program is loaded and executed by a processor, the method according to any one of claims 1 to 5 is adopted.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that: When the processor loads and executes the computer program, the method according to any one of claims 1 to 5 is adopted.

Images