An automatic synthesis method of step-by-step answer videos

Abstract

The application relates to a method and system for automatically synthesizing step-by-step answer videos, wherein when a question consultation request sent by a client is received by a server, current question information, N current step video relationship information corresponding to the current question information and a current answer video containing N current step video information are acquired according to the question consultation request, the current answer video is synthesized into a message and sent to the client, the client acquires a historical synthesized video according to the message, and the synthesis analysis results of four key frames of the historical synthesized video are matched with the message; if all the matching is successful, the historical synthesized video is directly played, otherwise, N video clips are acquired according to the N current step video information of the current answer video, and the N video clips are synthesized according to the N current step video relationship information in the client to obtain and play a current synthesized video. Therefore, the application reduces the resource loss of the server, relieves the pressure of the server and optimizes the user experience.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to an automatic method and system for synthesizing step-by-step answer videos. Background Technology

[0002] Traditional app-based customer service systems support basic intelligent reply functions and provide pre-recorded video answers for specific questions. When the question is more complex, they will return multiple video answers in steps, such as: see video A for step one, see video B for step two, see video C for step three, etc. Returning multiple video answers not only results in a poor user experience but also low response efficiency.

[0003] To optimize user experience and improve response efficiency, two common methods are: 1. Manually recording a complete answer video, including all steps, for specific complex questions; 2. Using technical means to merge video clips of each step into a single complete answer video. Regardless of the method, what is ultimately returned to the user is a complete answer video composed of individual step-by-step segments. This not only requires more server bandwidth and consumes more user data, but the server-side synthesis of the complete answer video also consumes both computing and storage resources, increasing server load. Summary of the Invention

[0004] The technical problem to be solved by this invention is: This invention provides an automatic synthesis method and system for step-by-step answer videos, which reduces server resource consumption, alleviates server pressure, optimizes user experience, and improves response efficiency.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0006] In a first aspect, the present invention provides an automatic synthesis method for step-by-step answer videos, comprising:

[0007] The server receives a question inquiry request sent by the client, retrieves the current question information, the N current step video relationship information corresponding to the current question information, and the current answer video containing the N current step video information from the database according to the question inquiry request, and combines the current question information, the N current step video relationship information, and the current answer video into a message and sends it to the client.

[0008] While receiving the message, the client obtains the historical composite video, reads the four key frames of the historical composite video, parses the second QR code of each key frame, obtains all composite parsing results containing historical issue ID, historical step video type and historical anti-duplicate identifier, matches all composite parsing results with the message, and if all matches are successful, the historical composite video is played directly. The four key frames are: the first frame of the composite, the last frame, the first intermediate frame and the second intermediate frame.

[0009] If the match fails, a historical video segment is retrieved, the first frame of the historical video segment is read, and the first QR code of the first frame of the segment is parsed to obtain a segment parsing result containing the video ID and video title. The segment parsing result is matched with the message. If the segment parsing result matches the message successfully, the video segment is synthesized on the client according to N video relationship information of the current step to obtain and play the current synthesized video.

[0010] The beneficial effects of this invention are as follows: When the server receives a client's inquiry request, unlike the traditional method of directly returning a complete answer video composed of fragments of each step to the client, the current question information, the relationship information of the N current step videos corresponding to the current question information, and the current answer video containing the N current step video information are synthesized into a message and sent to the client, improving transmission efficiency. The client uses a four-keyframe identification technique based on the received message, namely: synthesizing the first frame, the last frame, the first intermediate frame, and the second intermediate frame to determine whether there is a historical synthesized video. This can achieve fast identification while ensuring the integrity and accuracy of the identified historical synthesized video, avoiding unnecessary duplicate video synthesis, saving data, and optimizing user experience. If the historical synthesized video does not match the message, the video fragments are obtained and synthesized on the client side, reducing server resource consumption and alleviating server pressure.

[0011] Optionally, parsing the second QR code of each keyframe to obtain all synthesized parsing results, and matching all synthesized parsing results with the message includes:

[0012] When the keyframe is the first frame of the synthesis, the second QR code of the first frame of the synthesis is parsed to obtain a first synthesis parsing result containing the historical issue ID, the historical step video type and the historical anti-duplicate identifier. In the first synthesis parsing result, the historical step video type is a guide video and the historical anti-duplicate identifier is the text description corresponding to the first historical step video in the historical synthesis video.

[0013] When the keyframe is the last frame, the second QR code of the last frame is parsed to obtain a second composite parsing result containing the historical issue ID, the historical step video type and the historical anti-duplicate identifier. In the second composite parsing result, the historical step video type is the end video, and the historical anti-duplicate identifier is the text description corresponding to all historical step videos in the historical composite video and the video ID corresponding to all historical step videos.

[0014] When the keyframe is the first intermediate frame, the second QR code of the first intermediate frame is parsed to obtain a third composite parsing result containing the historical issue ID, the historical step video type, and the historical anti-duplicate identifier. In the third composite parsing result, the historical step video type is an intermediate video or a guide video, and the historical anti-duplicate identifier is the text description corresponding to the first historical step video in the historical composite video, or the video ID corresponding to the first historical step video in the historical composite video, or the text description corresponding to the second historical step video in the historical composite video.

[0015] When the keyframe is the second intermediate frame, the second QR code of the second intermediate frame is parsed to obtain a fourth composite parsing result containing the historical issue ID, the historical step video type, and the historical anti-duplicate identifier. In the fourth composite parsing result, the historical step video type is an intermediate video or a guide video, and the historical anti-duplicate identifier is the text description corresponding to the second historical step video in the historical composite video, or the video ID corresponding to the second historical step video in the historical composite video, or the text description corresponding to the third historical step video in the historical composite video.

[0016] The first, second, third, and fourth synthesized parsing results are matched with the message.

[0017] As described above, the historical step video type and historical anti-duplicate identifier differ for different keyframe synthesis analysis results. For the second synthesis analysis result where the keyframe is the tail frame, the historical anti-duplicate identifier consists of the text descriptions and video IDs corresponding to all historical step videos, ensuring the integrity and accuracy of the identified historical synthesis videos. For the third and fourth synthesis analysis results where the keyframes are the first and second intermediate frames, the historical step video type and historical anti-duplicate identifier have multiple possibilities. This diverse and flexible configuration fits the actual situation of answer videos composed of multiple steps, further improving the accuracy of the identified historical synthesis videos.

[0018] Optionally, matching the first synthesized parsing result, the second synthesized parsing result, the third synthesized parsing result, and the fourth synthesized parsing result with the message includes:

[0019] When matching the first synthetic parsing result with the message, the historical problem ID in the first synthetic parsing result is matched with the current problem ID in the current problem information of the message. If the match is successful, the text description corresponding to the first current step video is obtained from the N current step video relationship information and the N current step video information according to the current problem ID of the message. If the text description corresponding to the first current step video is consistent with the text description corresponding to the first historical step video in the historical synthetic video in the first synthetic parsing result, then the first synthetic parsing result is successfully matched with the message; otherwise, the first synthetic parsing result is not successfully matched with the message.

[0020] When matching the second synthetic parsing result with the message, the historical problem ID in the second synthetic parsing result is matched with the current problem ID in the current problem information of the message. If the match is successful, the text description and video ID corresponding to each current step video are obtained from the N current step video relationship information and the N current step video information according to the current problem ID of the message. If the text description and video ID corresponding to all current step videos are consistent with the text description and video ID corresponding to all historical step videos in the historical synthetic videos in the second synthetic parsing result, then the second synthetic parsing result is successfully matched with the message; otherwise, the second synthetic parsing result is unsuccessfully matched with the message.

[0021] When matching the third composite parsing result with the message, the historical problem ID in the third composite parsing result is matched with the current problem ID in the current problem information of the message. If the match is successful, the text description corresponding to the first historical step video, the video ID corresponding to the first historical step video, and the text description corresponding to the second historical step video are obtained from the N current step video relationship information and the N current step video information according to the current problem ID of the message. If the text description corresponding to the first historical step video is consistent with the text description corresponding to the first historical step video in the historical composite video in the third composite parsing result, or if the video ID corresponding to the first current step video is consistent with the video ID corresponding to the first historical step video in the historical composite video in the third composite parsing result, or if the text description corresponding to the second current step video is consistent with the text description corresponding to the second historical step video in the historical composite video in the third composite parsing result, then the third composite parsing result matches the message successfully; otherwise, the third composite parsing result does not match the message successfully.

[0022] When matching the fourth composite parsing result with the message, the historical problem ID in the fourth composite parsing result is matched with the current problem ID in the current problem information of the message. If the match is successful, the text description corresponding to the second current step video, the video ID corresponding to the second current step video, and the text description corresponding to the third current step video are obtained from the N current step video relationship information and the N current step video information based on the current problem ID of the message. If the text description corresponding to the second current step video is consistent with the text description corresponding to the second historical step video in the historical composite video in the fourth composite parsing result, or if the video ID corresponding to the second current step video is consistent with the video ID corresponding to the second historical step video in the historical composite video in the fourth composite parsing result, or if the text description corresponding to the third current step video is consistent with the text description corresponding to the third historical step video in the historical composite video in the fourth composite parsing result, then the fourth composite parsing result matches the message successfully; otherwise, the fourth composite parsing result does not match the message successfully.

[0023] As described above, when matching the first, second, third, and fourth composite parsing results with the message, each composite parsing result has its own matching method. All of them prioritize matching the historical issue ID with the current issue ID. Only when the match is successful will further judgment be made, thereby improving the accuracy and efficiency of identification and saving data.

[0024] Optionally, if the matching fails, obtaining N video segments based on the N current step video information of the current answer video, and synthesizing the N video segments on the client according to the N current step video relationship information includes:

[0025] If the match fails, retrieve the historical video segment, read the first frame of the historical video segment, and parse the first QR code of the first frame of the segment to obtain the segment parsing result containing the video ID and video title;

[0026] Obtain the video IDs and video titles corresponding to all current step videos from the N current step video relationship information and the N current step video information;

[0027] The segment parsing result is matched with the message. It is determined whether the video ID and video title in the segment parsing result contain the video ID and video title corresponding to all current step videos. If yes, the segment parsing result matches the message successfully. The historical video segments are then synthesized on the client according to N current step video relationship information. If no, the segment parsing result does not match the message successfully. P current step videos corresponding to the video ID and P current step videos corresponding to the video title are not included in the segment parsing result. P video segment links are obtained from the current answer video based on the P current step videos corresponding to the video ID and P current step videos corresponding to the video title. The corresponding P current video segments are downloaded from the Internet based on the P video segment links. The P current video segments and NP historical video segments are then synthesized on the client according to N current step video relationship information.

[0028] As described above, when merging video segments, historical video segments are first retrieved. The parsing results of these historical video segments are then matched with the message to determine if any historical video segments have been downloaded previously. If all historical video segments have been downloaded, the currently merged video is directly synthesized using the previously downloaded historical video segments, avoiding duplicate downloads and saving bandwidth while improving merging efficiency. If no historical video segments have been downloaded or not all historical video segments have been downloaded, the corresponding segment link is directly obtained from the current answer video in the message and downloaded to obtain the current video segment for merging, ensuring the integrity of the currently merged video and optimizing the user experience.

[0029] Optionally, the step of synthesizing the N video segments on the client according to the N current step video relationship information includes:

[0030] Obtain historical guide videos, read the first frame of the guide video, and parse the third QR code of the first frame to obtain a guide parsing result containing the current step number and the text description of the current step.

[0031] Obtain the current step sequence number and the current step text description corresponding to all current step videos from the N current step video relationship information and the N current step video information;

[0032] The wizard parsing result is matched with the message. It is determined whether the current step number and current step text description in the wizard parsing result contain the current step number and current step text description corresponding to all current step videos. If yes, the wizard parsing result matches the message successfully. Based on the N current step video information of the current answer video, the video segment under the current step number is combined with the historical wizard video according to the N current step video relationship information on the client. If no, the wizard parsing result does not match the message successfully. From the N current step video information of the current answer video, L current step video information corresponding to L current step videos that are not included in the wizard parsing result is obtained. Based on the L current step video information, the corresponding L current wizard videos are obtained. The L current wizard videos and NL historical wizard videos are combined with the video segment according to the N current step video relationship information on the client.

[0033] As described above, the process involves matching the parsing results of historical guide videos with the message. If all matches are successful, it indicates that historical guide videos have been downloaded previously. In this case, the historical guide videos and historical video clips are directly combined without repeated downloads, thus saving data. If none matches are successful, it indicates that historical guide videos were not downloaded or were not fully downloaded. In this case, the corresponding current guide video is downloaded, and the current guide video, historical guide videos, and video clips are combined to ensure that the resulting composite video includes both complete video clips and complete guide videos, thus guaranteeing the integrity of the composite video and optimizing the user experience.

[0034] Optionally, obtaining the corresponding L current guide videos based on the L current step video information includes:

[0035] Extract the background image link, video size, and text description of the current step from the L current step video information;

[0036] Download the corresponding background images from the Internet based on the background image links of the L current step videos to obtain L background images. Adjust the L background images according to the corresponding video size to obtain L adjusted background images. Use the L adjusted background images as the wizard screens of the corresponding current step videos to obtain L wizard screens.

[0037] The text descriptions of the current steps in the L current step videos are converted into corresponding L step audios using a TTS engine, resulting in L step audios.

[0038] Each wizard screen is synthesized with its corresponding step-by-step audio using FFmpeg, resulting in L current wizard videos.

[0039] As described above, the current step video information contains a link to the background image of the current step video, the video size, and a text description of the current step. This information is used to obtain the guide screen and the step audio, thereby achieving the synthesis of the current guide video and ensuring the accuracy of the synthesized current guide video.

[0040] Optionally, the step of synthesizing the L current guide videos and NL historical guide videos with the video clips according to N current step video relationship information on the client includes:

[0041] Obtain the current step number of each current step video from the N current step video relationship information, take the video size of the current step video with current step number 1 as the reference size, and adjust the L current guide videos, NL historical guide videos and video segments according to the reference size to obtain the adjusted L current guide videos, the adjusted NL historical guide videos and the adjusted video segments.

[0042] According to the current step number, the adjusted L current guide videos, the adjusted NL historical guide videos, and the adjusted video clips are synthesized on the client.

[0043] As described above, when compositing the current wizard video, historical wizard videos, and video clips, the video size of the current step video (current step number 1) is used as a reference size. This is used to adjust the video sizes of the current wizard video, historical wizard videos, and video clips, ensuring that the resulting composite video does not have inconsistent video sizes, thus guaranteeing the quality of the composite video and optimizing the user experience.

[0044] Optionally, the process of combining the adjusted L current wizard videos, the adjusted NL historical wizard videos, and the adjusted video clips on the client according to the current step number includes:

[0045] Using FFmpeg's quirc filter, the third QR code is located in L current guide videos and NL adjusted historical guide videos, while the first QR code is located in the adjusted video segments, thus obtaining the positions of the corresponding third QR code and the first QR code.

[0046] Based on the positions of the third and first QR codes, the delogo filter is used to remove the QR codes from the adjusted L current guide videos, the adjusted NL historical guide videos, and the adjusted video segments, respectively, resulting in the removed L current guide videos, the removed NL historical guide videos, and the removed video segments.

[0047] Obtain the video ID, video title, and current step text description from the N current step video information of the current answer video. Generate a new first QR code based on the video ID and video title. Generate a new third QR code based on the current step text description and the current step sequence number. Add the new first QR code to the cleared video segment according to the position of the first QR code. At the same time, add the new third QR code to the L cleared current guide videos and NL cleared historical guide videos according to the position of the third QR code. This results in the added video segment, the added L current guide videos, and the added NL historical guide videos.

[0048] According to the current step number, the added video clips, the L current guide videos, and the NL historical guide videos are combined on the client.

[0049] As described above, the QR codes of the current guide video, historical guide videos, and video clips are first cleared, and then new QR codes are generated and added again according to a dedicated QR code generation method. This ensures that the current guide video, historical guide videos, and video clips with the new QR codes are used in the final synthesis, guaranteeing the effectiveness of the generated current synthesized video and providing a basis for subsequent anti-duplicate measures.

[0050] In a second aspect, the present invention provides an automatic synthesis system for step-by-step answer videos, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the automatic synthesis method for step-by-step answer videos described in the first aspect.

[0051] The technical effects of the automatic synthesis system for step-by-step answer videos provided in the second aspect are the same as those of the automatic synthesis method for step-by-step answer videos provided in the first aspect. Attached Figure Description

[0052] Figure 1This is a flowchart illustrating an automatic synthesis method for step-by-step answer videos provided in this embodiment;

[0053] Figure 2 This is a schematic diagram of the overall process of an automatic synthesis method for step-by-step answer videos provided in this embodiment;

[0054] Figure 3 This is a schematic diagram of the structure of an automatic synthesis system for step-by-step answer videos provided in this embodiment.

[0055] [Explanation of Labels in the Attached Image]

[0056] 1. An automatic synthesis system for step-by-step answer videos;

[0057] 2. Processor;

[0058] 3. Memory. Detailed Implementation

[0059] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0060] Example 1

[0061] Please refer to Figures 1 to 2 This invention provides a multimodal image segmentation method, comprising the following steps:

[0062] S1. The server receives a question inquiry request sent by the client, retrieves the current question information, the N current step video relationship information corresponding to the current question information, and the current answer video containing the N current step video information from the database according to the question inquiry request, and combines the current question information, the N current step video relationship information and the current answer video into a message and sends it to the client.

[0063] In this embodiment, administrators pre-upload recorded video clips to the customer service management platform and fill in the video clip information, including but not limited to: video title, video content, video link, etc., and save them as video records to the database. Simultaneously, the video clip's video ID, creation time, and video dimensions (including video width and length) and duration are automatically generated from the video clip's metadata and saved to the video record. Furthermore, for each recorded video clip, administrators also add a first QR code to the lower right corner to prevent duplication. The first QR code consists of the video ID and video title. When adding complex questions to the customer service management platform, administrators fill in the question ID, question title, etc. The system generates a list of answer videos based on the question title and keyword list, selecting videos from the database in the order they were answered. Each answer step includes a text description, step number, background image link, and video dimensions, establishing a complete relationship between the current steps and the answer video. Additionally, administrators upload a corresponding guide video. This guide video also includes a QR code in the lower right corner to prevent duplication. The QR code consists of the step number and a text description, with wildcards inserted to allow for personalized descriptions for different users.

[0064] When a user asks a question on the customer service interface of the client app, the question categories will be listed first, guiding the user to select the question that best matches their needs. Then, a question inquiry request will be generated based on the question and sent to the server.

[0065] like Figure 2 As shown, the server receives a question request from the client, retrieves the current question information, the N current step video relationship information corresponding to the current question information, and the current answer video containing the N current step video information from the database based on the question request, and encapsulates the current question information, the N current step video relationship information, and the current answer video in a JSON message, that is, combines them into a message and sends it to the client.

[0066] In a specific embodiment, the specific message is as follows:

[0067]

[0068]

[0069] S2. While receiving the message, the client obtains the historical composite video, reads the four key frames of the historical composite video, parses the second QR code of each key frame, obtains all composite parsing results containing historical problem ID, historical step video type and historical anti-duplicate identifier, matches all composite parsing results with the message, and if all matches are successful, the historical composite video is played directly. The four key frames are: composite first frame, last frame, first intermediate frame and second intermediate frame.

[0070] In this embodiment, as Figure 2 As shown, the client receives the message and simultaneously retrieves the historical synthesized video. This historical synthesized video is a locally generated video on the client. It should be noted that when the client synthesizes a video locally, it adds a second QR code to the synthesized video. This second QR code consists of the question ID, video type, and anti-duplicate flag. The video type includes: guide video, intermediate video, and ending video. The anti-duplicate flag varies depending on the video type, as explained in subsequent steps. The client reads four keyframes of the historical synthesized video: the first frame, the last frame, the first intermediate frame, and the second intermediate frame. The second QR code of each keyframe is parsed to obtain all the synthesis parsing results containing the historical question ID, historical step video type, and historical anti-duplicate flag. All synthesis parsing results are matched against the message. If all matches are successful, it means that an answer video for the same question inquiry request has been synthesized previously, so the historical synthesized video is played directly.

[0071] At this point, step S2 involves parsing the second QR code of each keyframe to obtain all synthesized parsing results, and matching all synthesized parsing results with the message, including:

[0072] S21. When the key frame is the first frame of the synthesis, the second QR code of the first frame of the synthesis is parsed to obtain a first synthesis parsing result containing the historical problem ID, the historical step video type and the historical anti-duplicate identifier. In the first synthesis parsing result, the historical step video type is a guide video and the historical anti-duplicate identifier is the text description corresponding to the first historical step video in the historical synthesis video.

[0073] S22. When the keyframe is the last frame, the second QR code of the last frame is parsed to obtain a second composite parsing result containing the historical problem ID, the historical step video type and the historical anti-duplicate identifier. In the second composite parsing result, the historical step video type is the end video, and the historical anti-duplicate identifier is the text description corresponding to all historical step videos in the historical composite video and the video ID corresponding to all historical step videos.

[0074] S23. When the keyframe is the first intermediate frame, the second QR code of the first intermediate frame is parsed to obtain a third composite parsing result containing the historical issue ID, the historical step video type, and the historical anti-duplicate identifier. In the third composite parsing result, the historical step video type is an intermediate video or a guide video, and the historical anti-duplicate identifier is the text description corresponding to the first historical step video in the historical composite video, or the video ID corresponding to the first historical step video in the historical composite video, or the text description corresponding to the second historical step video in the historical composite video.

[0075] S24. When the keyframe is the second intermediate frame, the second QR code of the second intermediate frame is parsed to obtain a fourth composite parsing result containing the historical issue ID, the historical step video type, and the historical anti-duplicate identifier. In the fourth composite parsing result, the historical step video type is an intermediate video or a guide video, and the historical anti-duplicate identifier is the text description corresponding to the second historical step video in the historical composite video, or the video ID corresponding to the second historical step video in the historical composite video, or the text description corresponding to the third historical step video in the historical composite video.

[0076] S25. Match the first synthesized parsing result, the second synthesized parsing result, the third synthesized parsing result, and the fourth synthesized parsing result with the message.

[0077] In this embodiment, when the keyframe is the first frame of the synthesis, in the first synthesis parsing result containing the historical issue ID, historical step video type, and historical anti-duplicate identifier, the historical step video type is a guide video, and the historical anti-duplicate identifier is the text description corresponding to the first historical step video, which is the current step text description of the current step video. Thus, when the historical step video type is a guide video, its anti-duplicate identifier is generated based on the text description corresponding to the first historical step video. When the keyframe is the last frame, in the second synthesis parsing result containing the historical issue ID, historical step video type, and historical anti-duplicate identifier, the historical step video type is the end video, and the historical anti-duplicate identifier is the text description corresponding to all historical step videos in the historical synthesis video and the video ID corresponding to all historical step videos, that is, the current step text description of all current step videos and the video ID of all current step videos, which is also the video segment. Thus, when the historical step video type is the end video, its anti-duplicate identifier is generated based on the text description corresponding to all historical step videos and the video ID corresponding to all historical step videos. In fact, when N is less than or equal to 2, that is, when the current answer video in the obtained message is composed of less than or equal to 2 current step video information, it only synthesizes the first frame and the last frame, and there is no need to obtain and parse the subsequent first intermediate frame and the second intermediate frame. The following is an explanation for the case where N is greater than 2.

[0078] When the keyframe is the first intermediate frame, in the third-synthesized parsing result containing the historical question ID, historical step video type, and historical anti-duplicate identifier, the historical step video type may be an intermediate video or a guide video. The first intermediate frame is the frame at second X. The value of X is obtained by adding 1 to the video duration of the first current step video in the current answer video in the message, that is, the current step video with current step number 1. Considering that: when the length of the text description of the first current step video is greater than its video duration, the length of the text-to-speech / step-to-speech converted from the text description exceeds the value of X, therefore, the historical anti-duplicate identifier at this time is based on... The text description of the first current step video is generated, which is the text description corresponding to the first historical step video in the historical synthesized video. When the length of the text description of the first current step video is less than its video duration, the length of the text-to-speech / step-to-speech converted from the text description may not exceed the X value. Therefore, the historical anti-duplicate identifier at this time is generated based on the video ID of the first current step video, which is the video ID corresponding to the first historical step video in the historical synthesized video. It may also not exceed the X value. Then, the historical anti-duplicate identifier at this time is based on the text description of the second current step video, which is the text description corresponding to the second historical step video in the historical synthesized video.

[0079] When the keyframe is the second intermediate frame, in the fourth composite parsing result containing the historical question ID, historical step video type, and historical anti-duplicate identifier, the historical step video type may be either an intermediate video or a guide video. The second intermediate frame is the frame at second Y. The value of Y is obtained based on the video duration of the first current step video and the video duration of the second current step video in the current answer video in the message. That is, it is obtained by adding the video duration of the current step video with current step number 1 to the video duration of the current step video with current step number 2, taking into account that: when the text description length of the first current step video is greater than its video duration, or the text description length of the second current step video is greater than its video duration. When the sum of the converted text / speech / step speech and the video length of the first current step video exceeds the value of Y, the historical anti-duplicate identifier is generated based on the text description of the second current step video, i.e., the text description corresponding to the second historical step video in the historical synthesized video. When the sum of the converted text / speech / step speech and the video length of the first current step video does not exceed the value of Y, the historical anti-duplicate identifier is generated based on the video ID of the second current step video, i.e., the video ID corresponding to the second historical step video in the historical synthesized video, or based on the text description of the third current step video, i.e., the text description corresponding to the third historical step video in the historical synthesized video.

[0080] At this point, step S25 includes:

[0081] S251. When matching the first synthetic parsing result with the message, the historical problem ID in the first synthetic parsing result is matched with the current problem ID in the current problem information of the message. If the match is successful, the text description corresponding to the first current step video is obtained from N current step video relationship information and N current step video information according to the current problem ID of the message. If the text description corresponding to the first current step video is consistent with the text description corresponding to the first historical step video in the historical synthetic video in the first synthetic parsing result, then the first synthetic parsing result is successfully matched with the message; otherwise, the first synthetic parsing result is unsuccessfully matched with the message.

[0082] S252. When matching the second composite parsing result with the message, the historical problem ID in the second composite parsing result is matched with the current problem ID in the current problem information of the message. If the match is successful, the text description and video ID corresponding to each current step video are obtained from the N current step video relationship information and the N current step video information according to the current problem ID of the message. If the text description and video ID corresponding to all current step videos are consistent with the text description and video ID corresponding to all historical step videos in the historical composite videos in the second composite parsing result, then the second composite parsing result is successfully matched with the message; otherwise, the second composite parsing result is unsuccessfully matched with the message.

[0083] S253. When matching the third composite parsing result with the message, the historical problem ID in the third composite parsing result is matched with the current problem ID in the current problem information of the message. If the match is successful, the text description corresponding to the first historical step video, the video ID corresponding to the first historical step video, and the text description corresponding to the second historical step video are obtained from the N current step video relationship information and the N current step video information according to the current problem ID of the message. If the text description corresponding to the first historical step video is consistent with the text description corresponding to the first historical step video in the historical composite video in the third composite parsing result, or if the video ID corresponding to the first current step video is consistent with the video ID corresponding to the first historical step video in the historical composite video in the third composite parsing result, or if the text description corresponding to the second current step video is consistent with the text description corresponding to the second historical step video in the historical composite video in the third composite parsing result, then the third composite parsing result matches the message successfully; otherwise, the third composite parsing result does not match the message successfully.

[0084] S254. When matching the fourth composite parsing result with the message, the historical problem ID in the fourth composite parsing result is matched with the current problem ID in the current problem information of the message. If the match is successful, the text description corresponding to the second current step video, the video ID corresponding to the second current step video, and the text description corresponding to the third current step video are obtained from the N current step video relationship information and the N current step video information according to the current problem ID of the message. If the text description corresponding to the second current step video is consistent with the text description corresponding to the second historical step video in the historical composite video in the fourth composite parsing result, or if the video ID corresponding to the second current step video is consistent with the video ID corresponding to the second historical step video in the historical composite video in the fourth composite parsing result, or if the text description corresponding to the third current step video is consistent with the text description corresponding to the third historical step video in the historical composite video in the fourth composite parsing result, then the fourth composite parsing result matches the message successfully; otherwise, the fourth composite parsing result does not match the message successfully.

[0085] As described above, when matching the first, second, third, and fourth composite parsing results with the message, the first composite parsing result is prioritized. If the historical issue ID in the first composite parsing result does not match the current issue ID in the message's current issue information, the next historical composite video is directly retrieved. Only when the historical issue ID in the first composite parsing result matches the current issue ID in the message's current issue information will the second composite parsing result be matched with the message. Similarly, if the second composite parsing result fails to match the message, the next historical composite video is directly retrieved. Only when all historical steps in the second composite parsing result match... The text description corresponding to the step video and the video ID corresponding to all historical step videos must match the text description corresponding to the current step video and the video ID corresponding to all current step videos in the message. That is, the third synthesis parsing result will be matched with the message only when the match is successful. Moreover, when matching the third synthesis parsing result with the message, multiple cases of the third synthesis parsing result are taken into account. Similarly, if the third synthesis parsing result does not match the message, the next historical synthesis video is directly obtained. The fourth synthesis parsing result will be matched with the message only when the third synthesis parsing result matches the message successfully. Moreover, when matching the fourth synthesis parsing result with the message, multiple cases of the fourth synthesis parsing result are taken into account.

[0086] S3. If the matching fails, obtain N video segments based on the N current step video information of the current answer video, synthesize the N video segments on the client according to the N current step video relationship information, and obtain and play the current synthesized video.

[0087] In this embodiment, as Figure 2 As shown, if the historical synthesized video does not match the message, that is, the current synthesized video has not been synthesized before, then N video segments are downloaded from the Internet according to the video links corresponding to the N current step video information of the current answer video. The N video segments are synthesized on the client according to the N current step video relationship information to obtain and play the current synthesized video.

[0088] At this point, if the matching fails in step S3, then N video segments are obtained based on the N current step video information of the current answer video, and the N video segments are synthesized according to the N current step video relationship information, including:

[0089] S31. If the matching fails, obtain the historical video segment, read the first frame of the historical video segment, and obtain the segment parsing result containing the video ID and video title by parsing the first QR code of the first frame of the segment.

[0090] S32. Obtain the video IDs and video titles corresponding to all current step videos from the N current step video relationship information and the N current step video information;

[0091] S33. Match the segment parsing result with the message, and determine whether the video ID and video title in the segment parsing result contain the video ID and video title corresponding to all current step videos. If yes, the segment parsing result matches the message successfully, and the historical video segments are synthesized on the client according to N current step video relationship information. If no, the segment parsing result does not match the message successfully, obtain the P current step videos corresponding to the video IDs and P current step videos corresponding to the video titles that are not included in the segment parsing result, obtain P video segment links from the current answer video according to the P current step videos corresponding to the video IDs and P current step videos corresponding to the video titles, download the corresponding P current video segments from the Internet according to the P video segment links, and synthesize the P current video segments and NP historical video segments on the client according to N current step video relationship information.

[0092] In this embodiment, as Figure 2As shown, to obtain historical video clips, the process involves reading the first frame of a historical video clip from the client's local storage, parsing the first QR code in the first frame to obtain a clip parsing result containing the video ID and video title. From the N current-step video relationship information and N current-step video information in the message, the video IDs and titles corresponding to all current-step videos are retrieved. The clip parsing result is then matched with the message to determine if the video IDs and titles in the clip parsing result contain all the video IDs and titles corresponding to all current-step videos. If they do, it means that the video clip required for the current answer to this question / consultation request has been downloaded previously, and the historical video clip can be directly downloaded. The video segments are synthesized according to the N current step video relationship information. Conversely, if the segment parsing result does not contain the video IDs and video titles corresponding to all current step videos, then the corresponding video segments need to be obtained. P current step video IDs and P current step video titles that are not included in the segment parsing result are obtained. Based on the P current step video IDs and P current step video titles, P video segment links are obtained from the current answer video. Based on the P video segment links, the corresponding P current video segments are downloaded from the Internet. The P current video segments and NP historical video segments are synthesized according to the N current step video relationship information. At this time, P may be equal to N.

[0093] At this point, step S3, which involves synthesizing the N video segments according to the N current step video relationship information, includes:

[0094] S34. Obtain historical guide videos, read the first frame of the guide video, and obtain a guide parsing result containing the current step number and the text description of the current step by parsing the third QR code of the first frame of the guide.

[0095] S35. Obtain the current step sequence number and the current step text description corresponding to all current step videos from the N current step video relationship information and the N current step video information;

[0096] S36. Match the wizard parsing result with the message, and determine whether the current step number and current step text description in the wizard parsing result contain the current step number and current step text description corresponding to all current step videos. If yes, the wizard parsing result matches the message successfully. Based on the N current step video information of the current answer video, synthesize the video segment under the current step number with the historical wizard video according to the N current step video relationship information on the client. If no, the wizard parsing result does not match the message successfully. Obtain L current step video information corresponding to L current step videos that are not included in the wizard parsing result from the N current step video information of the current answer video. Obtain the corresponding L current wizard videos based on the L current step video information. Synthesize the L current wizard videos and NL historical wizard videos with the video segment according to the N current step video relationship information on the client.

[0097] In this embodiment, as Figure 2 As shown, the process involves retrieving historical guide videos, reading the first frame of the guide video, parsing the third QR code in the first frame to obtain a guide parsing result containing the current step number and text description. From the N current step video relationship information in the message and the N current step video information in the current answer video, the process retrieves the current step number and text description corresponding to all current step videos. The guide parsing result is then matched with the message. If the current step number and text description in the guide parsing result contain the current step number and text description corresponding to all current step videos, it indicates that the client previously downloaded all the guide videos required for the current answer video of the question inquiry request. Therefore, the process can be directly performed based on the N current step video relationships. The previous step video information is synthesized by combining the video segment under the current step number with the corresponding historical guide video according to the N current step video relationship information. If the current step number and current step text description in the guide parsing result do not contain the current step number and current step text description corresponding to all current step videos, it means that the client has not downloaded or has not fully downloaded the guide video before. Therefore, from the N current step video information of the current answer video, L current step video information corresponding to L current step videos that are not included in the guide parsing result is obtained. Based on the L current step video information, the corresponding L current guide videos are obtained. The L current guide videos and NL historical guide videos are synthesized with the video segment according to the N current step video relationship information. At this time, L may be equal to N.

[0098] At this point, step S36, which involves obtaining the corresponding L current guide videos based on the L current step video information, includes:

[0099] S361. Obtain the background image link, video size, and text description of the current step from the L current step video information;

[0100] S362. Download the corresponding background images from the Internet based on the background image links of the L current step videos to obtain L background images, and adjust the L background images according to the corresponding video size to obtain L adjusted background images. Use the L adjusted background images as the wizard screens of the corresponding current step videos to obtain L wizard screens.

[0101] S363. Use the TTS engine to convert the text descriptions of the current steps corresponding to the L current step videos into the corresponding L step audio, and obtain the L step audio.

[0102] S364. Use FFmpeg to synthesize each wizard screen with the corresponding step audio to obtain L current wizard videos.

[0103] In this embodiment, the background image links, video dimensions, and text descriptions of the current steps are obtained from L current step videos. The video dimensions include video length and video width. The corresponding background images are downloaded from the Internet according to the background image links to obtain L background images. The L background images are then adjusted according to the corresponding video dimensions, that is, so that the generated guide video has the same video dimensions as its corresponding video segment / current step video. The L adjusted background images are used as guide screens for the corresponding current step videos to obtain L guide screens. The text descriptions of the current steps corresponding to the L current step videos are converted into corresponding L step audio using a TTS engine to obtain L step audio. Each guide screen and its corresponding step audio are then synthesized using FFmpeg to obtain L current guide videos.

[0104] At this point, step S36, which involves combining the L current guide videos and NL historical guide videos with the video clips according to the N current step video relationship information, includes the following:

[0105] S365. Obtain the current step number of each current step video from the N current step video relationship information, take the video size of the current step video with current step number 1 as the reference size, and adjust the L current guide videos, NL historical guide videos and video segments according to the reference size to obtain the adjusted L current guide videos, the adjusted NL historical guide videos and the adjusted video segments.

[0106] S366. According to the current step number, the adjusted L current guide videos, the adjusted NL historical guide videos, and the adjusted video clips are synthesized on the client.

[0107] In this embodiment, the video size of the current step video with current step number 1 is used as the reference size. The L current wizard videos, NL historical wizard videos and video segments are adjusted according to the reference size so that all current wizard videos, all historical wizard videos and video segments have the same video size. Then, the adjusted L current wizard videos, the adjusted NL historical wizard videos and the adjusted video segments are combined.

[0108] At this point, step S366 includes:

[0109] S3661. Using FFmpeg's quirc filter, locate the third QR code for L current guide videos and NL adjusted historical guide videos respectively, and locate the first QR code for the adjusted video segments to obtain the corresponding third QR code position and first QR code position.

[0110] S3662. Based on the positions of the third QR code and the first QR code, the delogo filter is used to remove the QR codes from the adjusted L current guide videos, the adjusted NL historical guide videos, and the adjusted video segments, respectively, to obtain the removed L current guide videos, the removed NL historical guide videos, and the removed video segments.

[0111] S3663. Obtain the video ID, video title, and current step text description from the N current step video information of the current answer video. Generate a new first QR code based on the video ID and video title. Generate a new third QR code based on the current step text description and the current step number. Add the new first QR code to the cleared video segment according to the position of the first QR code. At the same time, add the new third QR code to the L cleared current guide videos and NL cleared historical guide videos according to the position of the third QR code. This results in the added video segment, the added L current guide videos, and the added NL historical guide videos.

[0112] S3664. According to the current step number, the added video clips, the L current guide videos, and the NL historical guide videos are combined on the client.

[0113] In this embodiment, the third QR code is located in L current guide videos and NL adjusted historical guide videos using FFmpeg's quirc filter. Simultaneously, the first QR code is located in the adjusted video segments. This yields the positions of the corresponding third and first QR codes. Based on these positions, the QR codes are cleared. The video ID, video title, and current step description are obtained from the N current step videos of the current answer video. A new first QR code is generated based on the video ID and video title. A new third QR code is generated based on the current step description and current step number. The new first and third QR codes are then re-added to their respective video segments. Simultaneously, the new third QR codes are re-added to the cleared L current guide videos and NL cleared historical guide videos according to their positions. Finally, the added video segments, the added L current guide videos, and the added NL historical guide videos are synthesized on the client side.

[0114] Example 2

[0115] Please refer to Figure 3 The present invention provides an automatic synthesis system 1 for step-by-step answer videos, including a memory 3, a processor 2, and a computer program stored in the memory 3 and executable on the processor 2. When the processor 2 executes the computer program, it implements the steps in Embodiment 1.

[0116] Since the systems / devices described in the above embodiments of the present invention are systems / devices used to implement the methods of the above embodiments of the present invention, those skilled in the art can understand the specific structure and modifications of the systems / devices based on the methods described in the above embodiments of the present invention, and therefore will not be repeated here. All systems / devices used in the methods of the above embodiments of the present invention fall within the scope of protection of the present invention.

[0117] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0118] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0119] It should be noted that any reference numerals placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims that enumerate several means, several of these means may be embodied by the same hardware. The use of the terms first, second, third, etc., is merely for convenience of expression and does not indicate any order. These terms can be understood as part of the component names.

[0120] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0121] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the claims should be interpreted to include both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0122] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, then this invention should also include these modifications and variations.

Claims

1. A method for automatically synthesizing step-by-step answer videos, characterized in that, include: The server receives a question inquiry request sent by the client, retrieves the current question information, the N current step video relationship information corresponding to the current question information, and the current answer video containing the N current step video information from the database according to the question inquiry request, and combines the current question information, the N current step video relationship information, and the current answer video into a message and sends it to the client. While receiving the message, the client obtains the historical composite video, reads the four key frames of the historical composite video, parses the second QR code of each key frame, obtains all composite parsing results containing historical issue ID, historical step video type and historical anti-duplicate identifier, matches all composite parsing results with the message, and if all matches are successful, the historical composite video is played directly. The four key frames are: the first frame of the composite, the last frame, the first intermediate frame and the second intermediate frame. If the matching fails, N video segments are obtained based on the N current step video information of the current answer video. The N video segments are then synthesized on the client according to the N current step video relationship information to obtain and play the current synthesized video. The step of parsing the second QR code of each keyframe to obtain all synthesized parsing results, and matching all synthesized parsing results with the message, includes: When the keyframe is the first frame of the synthesis, the second QR code of the first frame of the synthesis is parsed to obtain a first synthesis parsing result containing the historical issue ID, the historical step video type and the historical anti-duplicate identifier. In the first synthesis parsing result, the historical step video type is a guide video and the historical anti-duplicate identifier is the text description corresponding to the first historical step video in the historical synthesis video. When the keyframe is the last frame, the second QR code of the last frame is parsed to obtain a second composite parsing result containing the historical issue ID, the historical step video type and the historical anti-duplicate identifier. In the second composite parsing result, the historical step video type is the end video, and the historical anti-duplicate identifier is the text description corresponding to all historical step videos in the historical composite video and the video ID corresponding to all historical step videos. When the keyframe is the first intermediate frame, the second QR code of the first intermediate frame is parsed to obtain a third composite parsing result containing the historical issue ID, the historical step video type, and the historical anti-duplicate identifier. In the third composite parsing result, the historical step video type is an intermediate video or a guide video, and the historical anti-duplicate identifier is the text description corresponding to the first historical step video in the historical composite video, or the video ID corresponding to the first historical step video in the historical composite video, or the text description corresponding to the second historical step video in the historical composite video. When the keyframe is the second intermediate frame, the second QR code of the second intermediate frame is parsed to obtain a fourth composite parsing result containing the historical issue ID, the historical step video type, and the historical anti-duplicate identifier. In the fourth composite parsing result, the historical step video type is an intermediate video or a guide video, and the historical anti-duplicate identifier is the text description corresponding to the second historical step video in the historical composite video, or the video ID corresponding to the second historical step video in the historical composite video, or the text description corresponding to the third historical step video in the historical composite video. The first, second, third, and fourth synthesized parsing results are matched with the message.

2. The automatic synthesis method for step-by-step answer videos as described in claim 1, characterized in that, The step of matching the first synthesized parsing result, the second synthesized parsing result, the third synthesized parsing result, and the fourth synthesized parsing result with the message includes: When matching the first synthetic parsing result with the message, the historical problem ID in the first synthetic parsing result is matched with the current problem ID in the current problem information of the message. If the match is successful, the text description corresponding to the first current step video is obtained from the N current step video relationship information and the N current step video information according to the current problem ID of the message. If the text description corresponding to the first current step video is consistent with the text description corresponding to the first historical step video in the historical synthetic video in the first synthetic parsing result, then the first synthetic parsing result is successfully matched with the message; otherwise, the first synthetic parsing result is not successfully matched with the message. When matching the second synthetic parsing result with the message, the historical problem ID in the second synthetic parsing result is matched with the current problem ID in the current problem information of the message. If the match is successful, the text description and video ID corresponding to each current step video are obtained from the N current step video relationship information and the N current step video information according to the current problem ID of the message. If the text description and video ID corresponding to all current step videos are consistent with the text description and video ID corresponding to all historical step videos in the historical synthetic videos in the second synthetic parsing result, then the second synthetic parsing result is successfully matched with the message; otherwise, the second synthetic parsing result is unsuccessfully matched with the message. When matching the third composite parsing result with the message, the historical problem ID in the third composite parsing result is matched with the current problem ID in the current problem information of the message. If the match is successful, the text description corresponding to the first historical step video, the video ID corresponding to the first historical step video, and the text description corresponding to the second historical step video are obtained from the N current step video relationship information and the N current step video information according to the current problem ID of the message. If the text description corresponding to the first historical step video is consistent with the text description corresponding to the first historical step video in the historical composite video in the third composite parsing result, or if the video ID corresponding to the first current step video is consistent with the video ID corresponding to the first historical step video in the historical composite video in the third composite parsing result, or if the text description corresponding to the second current step video is consistent with the text description corresponding to the second historical step video in the historical composite video in the third composite parsing result, then the third composite parsing result matches the message successfully; otherwise, the third composite parsing result does not match the message successfully. When matching the fourth composite parsing result with the message, the historical problem ID in the fourth composite parsing result is matched with the current problem ID in the current problem information of the message. If the match is successful, the text description corresponding to the second current step video, the video ID corresponding to the second current step video, and the text description corresponding to the third current step video are obtained from the N current step video relationship information and the N current step video information based on the current problem ID of the message. If the text description corresponding to the second current step video is consistent with the text description corresponding to the second historical step video in the historical composite video in the fourth composite parsing result, or if the video ID corresponding to the second current step video is consistent with the video ID corresponding to the second historical step video in the historical composite video in the fourth composite parsing result, or if the text description corresponding to the third current step video is consistent with the text description corresponding to the third historical step video in the historical composite video in the fourth composite parsing result, then the fourth composite parsing result matches the message successfully; otherwise, the fourth composite parsing result does not match the message successfully.

3. The automatic synthesis method for step-by-step answer videos as described in claim 1, characterized in that, If the matching fails, then N video segments are obtained based on the N current step video information of the current answer video, and the N video segments are synthesized on the client according to the N current step video relationship information, including: If the match fails, retrieve the historical video segment, read the first frame of the historical video segment, and parse the first QR code of the first frame of the segment to obtain the segment parsing result containing the video ID and video title; Obtain the video IDs and video titles corresponding to all current step videos from the N current step video relationship information and the N current step video information; The segment parsing result is matched with the message. It is determined whether the video ID and video title in the segment parsing result contain the video ID and video title corresponding to all current step videos. If yes, the segment parsing result matches the message successfully. The historical video segments are then synthesized on the client according to N current step video relationship information. If no, the segment parsing result does not match the message successfully. P current step videos corresponding to the video ID and P current step videos corresponding to the video title are not included in the segment parsing result. P video segment links are obtained from the current answer video based on the P current step videos corresponding to the video ID and P current step videos corresponding to the video title. The corresponding P current video segments are downloaded from the Internet based on the P video segment links. The P current video segments and NP historical video segments are then synthesized on the client according to N current step video relationship information.

4. The automatic synthesis method for step-by-step answer videos as described in claim 1, characterized in that, The step of synthesizing the N video segments on the client according to the N current step video relationship information includes: Obtain historical guide videos, read the first frame of the guide video, and parse the third QR code of the first frame to obtain a guide parsing result containing the current step number and the text description of the current step. Obtain the current step sequence number and the current step text description corresponding to all current step videos from the N current step video relationship information and the N current step video information; The wizard parsing result is matched with the message. It is determined whether the current step number and current step text description in the wizard parsing result contain the current step number and current step text description corresponding to all current step videos. If yes, the wizard parsing result matches the message successfully. Based on the N current step video information of the current answer video, the video segment under the current step number is combined with the historical wizard video according to the N current step video relationship information on the client. If no, the wizard parsing result does not match the message successfully. From the N current step video information of the current answer video, L current step video information corresponding to L current step videos that are not included in the wizard parsing result is obtained. Based on the L current step video information, the corresponding L current wizard videos are obtained. The L current wizard videos and NL historical wizard videos are combined with the video segment according to the N current step video relationship information on the client.

5. The automatic synthesis method for step-by-step answer videos as described in claim 4, characterized in that, The step of obtaining the corresponding L current guide videos based on L current step video information includes: Extract the background image link, video size, and text description of the current step from the L current step video information; Download the corresponding background images from the Internet based on the background image links of the L current step videos to obtain L background images. Adjust the L background images according to the corresponding video size to obtain L adjusted background images. Use the L adjusted background images as the wizard screens of the corresponding current step videos to obtain L wizard screens. The text descriptions of the current steps in the L current step videos are converted into corresponding L step audios using a TTS engine, resulting in L step audios. Each wizard screen is synthesized with its corresponding step-by-step audio using FFmpeg, resulting in L current wizard videos.

6. The automatic synthesis method for step-by-step answer videos as described in claim 4, characterized in that, The step of synthesizing L current guide videos and NL historical guide videos with the video clips on the client according to N current step video relationship information includes: Obtain the current step number of each current step video from the N current step video relationship information, take the video size of the current step video with current step number 1 as the reference size, and adjust the L current guide videos, NL historical guide videos and video segments according to the reference size to obtain the adjusted L current guide videos, the adjusted NL historical guide videos and the adjusted video segments. According to the current step number, the adjusted L current guide videos, the adjusted NL historical guide videos, and the adjusted video clips are synthesized on the client.

7. The automatic synthesis method for step-by-step answer videos as described in claim 6, characterized in that, According to the current step number, the adjusted L current guide videos, the adjusted NL historical guide videos, and the adjusted video clips are synthesized on the client, including: Using FFmpeg's quirc filter, the third QR code is located in L current guide videos and NL adjusted historical guide videos, while the first QR code is located in the adjusted video segments, thus obtaining the positions of the corresponding third QR code and the first QR code. Based on the positions of the third and first QR codes, the delogo filter is used to remove the QR codes from the adjusted L current guide videos, the adjusted NL historical guide videos, and the adjusted video segments, respectively, resulting in the removed L current guide videos, the removed NL historical guide videos, and the removed video segments. Obtain the video ID, video title, and current step text description from the N current step video information of the current answer video. Generate a new first QR code based on the video ID and video title. Generate a new third QR code based on the current step text description and the current step sequence number. Add the new first QR code to the cleared video segment according to the position of the first QR code. At the same time, add the new third QR code to the L cleared current guide videos and NL cleared historical guide videos according to the position of the third QR code. This results in the added video segment, the added L current guide videos, and the added NL historical guide videos. According to the current step number, the added video clips, the L current guide videos, and the NL historical guide videos are combined on the client.

8. An automatic synthesis system for step-by-step answer videos, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 7.

Images