Slide video automatic generation method and device, equipment and medium

Abstract

The invention relates to the technical field of video generation, can be applied to the fields of financial science and technology and medical health, and discloses a slide video automatic generation method, device, equipment and medium, and the method comprises the steps: obtaining a slide file which comprises a plurality of slides and explanation information corresponding to each slide; according to the explanation information and input target video description information, text content in the slide is processed to obtain key text information, and the key text information is converted into paraphrasing voice; non-text content in the slide is extracted through a computer vision technology, the non-text content is processed to obtain key non-text information, and the key non-text information in the slide is highlighted to obtain a highlighted display slide; and generating a target video through a preset pseudo-autoregression model according to the parastyle content corresponding to the nearby voice and the highlighted display slide. And the generated video quality and the generation efficiency are effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of video generation technology and can be applied to the fields of financial technology and medical health. In particular, it relates to a method, device, equipment and medium for automatically generating slide videos. Background Art

[0002] Currently, in the fields of financial technology and healthcare, generating videos with narration and synchronized highlights is an important demand (for example, in the field of financial technology, the dynamic highlighting of K-line charts is synchronized with the voice commentary, and in the field of healthcare, the CT scan layer is synchronized with the pathological explanation). The generation of videos in the existing technology is mainly achieved through the following methods: 1. Relying on manual editing of videos, synchronizing the slide content with the narration, and manually adding highlight effects, which is not only time-consuming and labor-intensive, but also inefficient, and prone to errors, making it difficult to ensure the quality and consistency of the video; 2. Automatically generating lecture videos, but can only process simple slide content, and it is difficult to process complex text and image content. When processing multiple pages of slides and long narrations, the quality of the generated video is poor, which results in the generated video being visually unnatural and the user experience being poor. Summary of the Invention

[0003] The present invention provides a method, device, computer equipment and medium for automatically generating slide videos, so as to solve the technical problem of poor quality of existing slide video generation.

[0004] In a first aspect, a method for automatically generating a slideshow video is provided, comprising:

[0005] Obtaining a slide file, wherein the slide file includes multiple slides and explanation information corresponding to each slide;

[0006] Processing the text content in the slideshow according to the explanation information and the input target video description information to obtain key text information, and converting the key text information into narration voice;

[0007] extracting non-text content from the slide using computer vision technology, processing the non-text content to obtain key non-text information, and highlighting the key non-text information in the slide to obtain a highlighted slide;

[0008] A target video is generated according to the narration content corresponding to the side voice and the highlighted slides through a preset pseudo-autoregressive model.

[0009] In a second aspect, a device for automatically generating a slideshow video is provided, comprising:

[0010] An acquiring unit, configured to acquire a slide file, wherein the slide file includes a plurality of slides and explanation information corresponding to each of the slides;

[0011] a processing and conversion unit, configured to process the text content in the slide according to the explanation information and the input target video description information to obtain key text information, and convert the key text information into narration voice;

[0012] an extraction and display unit, configured to extract non-text content from the slide using computer vision technology, process the non-text content to obtain key non-text information, and highlight the key non-text information in the slide to obtain a highlighted slide;

[0013] A generating unit is used to generate a target video according to the narration content corresponding to the side voice and the highlighted slides through a preset pseudo autoregressive model.

[0014] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned method for automatically generating a slide video when executing the computer program.

[0015] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned method for automatically generating a slide video are implemented.

[0016] In the scheme implemented by the above-mentioned automatic generation method, device, computer equipment and storage medium of slide video, a slide file can be obtained, wherein the slide file includes multiple slides and explanation information corresponding to each slide; the text content in the slide is processed according to the explanation information and the input target video description information to obtain key text information, and the key text information is converted into narration voice; the non-text content in the slide is extracted through computer vision technology, and the non-text content is processed to obtain key non-text information, and the key non-text information in the slide is highlighted to obtain a highlighted slide; the target video is generated according to the narration content corresponding to the narration and the highlighted slide through a preset pseudo-autoregressive model. In the present invention, key text information in a slide is first converted into narration speech, and the key text information is obtained by fusing multi-source information; then, the non-text content in the slide is processed to obtain key non-text information, and the key non-text information in the slide is highlighted to obtain a highlighted slide; finally, the narration content and the highlighted slide corresponding to the narration speech are input into a preset pseudo-autoregressive model to generate a target video. During the target video generation process, the highlighted slide is dynamically adjusted according to the narration content through the pseudo-autoregressive model, so that the narration content is aligned with the adjusted highlighted slide, effectively improving the generated video quality and generation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0018] Figure 1 This is a flow chart of a method for automatically generating a slideshow video in one embodiment of the present invention;

[0019] Figure 2 yes Figure 1 A flow chart of a specific implementation of step S120;

[0020] Figure 3 yes Figure 1 A schematic flow chart of a specific implementation of step S130;

[0021] Figure 4 yes Figure 1 A schematic flow chart of a specific implementation of step S140;

[0022] Figure 5is a schematic block diagram of a device for automatically generating slideshow videos in one embodiment of the present invention;

[0023] Figure 6 is a structural diagram of a computer device in one embodiment of the present invention;

[0024] Figure 7 FIG. 2 is another structural diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0026] The method for automatically generating slide videos provided in an embodiment of the present invention can be applied to a client or a server, wherein the client can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, and portable wearable devices. The server can be implemented with an independent server or a server cluster consisting of multiple servers. At present, in the fields of financial technology and medical health, the quality of slide video generation is poor. In response to the above problems, the present invention proposes a method for automatically generating slide videos. The method can dynamically adjust the highlighted slides through a pseudo-autoregressive model according to the narration content, effectively improving the quality and generation efficiency of the generated video. The present invention is described in detail below through specific embodiments.

[0027] See also Figure 1 As shown, Figure 1 A flowchart of a method for automatically generating a slide video according to an embodiment of the present invention includes the following steps: S110-S140.

[0028] S110: Obtain a slide file, wherein the slide file includes multiple slides and explanation information corresponding to each slide.

[0029] Specifically, a slide file is obtained, wherein the slide file includes multiple slides and explanation information corresponding to each slide, and is usually stored in PPT (PowerPoint), PPTX or other similar presentation formats, and contains not only the visual content of each slide (such as text, charts, pictures, etc.), but also explanation information related to each slide. It should be noted that a slide refers to a single page in a slide file, which is usually used to display a specific theme or knowledge point. Each slide may contain elements such as text, charts, pictures, and graphics to convey specific information and display the content of a certain theme. The explanation information refers to the text content associated with each slide in the slide file, which is used to assist in explaining or illustrating the content of the slide, including detailed text descriptions, key points of explanation, supplementary information, etc.

[0030] In one possible implementation, the slide file may be a file of K-line chart highlights and risk voice explanations in the field of financial technology, or a file of CT scan layers and pathology explanations in the field of medical health.

[0031] S120 , processing the text content in the slide according to the explanation information and the input target video description information to obtain key text information, and converting the key text information into narration voice.

[0032] Specifically, for each slide, the text content in the slide is processed based on the slide's explanation information and the input target video description information to obtain key text information, where the target video description information is the rhythm plan of the target video (e.g., "The core formula should be highlighted in the first 30 seconds"). It can be understood that processing the text content in the slide based on the explanation information and the target video description information achieves the fusion of multi-source information, allowing for richer information expression. The key text information is then converted into voiceover narration.

[0033] Among them, such as Figure 2 As shown, step S120 includes the following steps: S121-S126:

[0034] S121, obtaining the target video description information;

[0035] S122: Inputting the text content in the slide, the target video description information, and the explanation information into a key information extraction model to extract key information from the text content, wherein the key information extraction model is a model based on artificial intelligence and natural language processing technology;

[0036] S123, marking the position of the key information in the slide to obtain the key text information;

[0037] S124, detecting whether target timbre information is input;

[0038] S125: if the target timbre information is input, converting the target timbre information and the key text information into the narration voice using TTS technology;

[0039] S126: If the target timbre information is not input, then obtaining default timbre information, and converting the default timbre information and the key text information into the narration voice using the TTS technology.

[0040] Specifically, the key information extraction model is a model based on artificial intelligence and natural language processing technologies, used to extract key information from text, that is, to automatically identify important information in slides. It is usually obtained based on deep learning algorithms (such as the BERT model, Transformer, etc.) or natural language processing technologies (such as NLP technology). It should be noted that the Transformer model is a neural network based on the self-attention mechanism, capable of processing long sequences of data, and is suitable for text processing and speech synthesis tasks. A suitable deep learning architecture (such as the Transformer) is selected, and the labeled data is input into the model for training. The model parameters are adjusted using an optimization algorithm (such as Adam or SGD) to minimize the error between the predicted results and the true annotations, thereby obtaining a key information identification model. During the training process, data augmentation, transfer learning, and multi-task learning techniques are used to improve the model's generalization ability and recognition accuracy. It should also be noted that the key information extraction model can integrate explanation annotation tags, match the target video description requirements (if the description requires "emphasis on contrasting data" → locate the difference value text), and perform contextual redundancy removal, removing repeated auxiliary statements and retaining unique core propositions (such as theorem definitions). More specifically, when converting the key text information into voice narration, it first checks whether the target timbre information has been entered. If so, the target timbre information and the key text information are converted into the voice narration using TTS technology. If not, the default timbre information and the key text information are converted into the voice narration using TTS technology. It should be noted that TTS (Text-to-Speech) technology can convert text content into voice output, generating natural and fluent speech by analyzing the semantic and grammatical structure of the text.

[0041] S130 , extracting non-text content from the slide using computer vision technology, processing the non-text content to obtain key non-text information, and highlighting the key non-text information in the slide to obtain a highlighted slide.

[0042] Specifically, the computer vision technology includes OCR technology and convolutional neural networks, and the non-text content includes formula content, chart content, and image content. It should be noted that in this embodiment, OCR (Optical Character Recognition) captures paper document images using an optical scanning device (such as a scanner or digital camera), analyzes the text morphological features using a pattern recognition algorithm, and converts them into editable computer text. The convolutional neural network is used to process image data and, in the present invention, is used to extract non-text content from the slide. It should also be noted that the key non-text information in the slide is highlighted to obtain a highlighted slide. Specifically, the key non-text information is highlighted. If it is formula content, a luminous animation (0.3s fade-in) is generated along the LaTeX symbol outline. If it is chart content, a focus mask (dynamically magnified from 80% to 100%) is superimposed on the data points. If it is image content, a semi-transparent marker layer (such as a red overlay for heart valves) is drawn in the semantic label area, and the highlighted layer is then alpha-blended with the original slide (with a transparency of 60%), finally outputting a highlighted slide with spatial positioning.

[0043] Among them, such as Figure 3 As shown, step S130 includes the following steps: S131-S137:

[0044] S131, scanning the slide using the computer vision technology to determine the boundary coordinates of the non-text content;

[0045] S132, classifying the bounding boxes corresponding to the bounding coordinates according to visual features to obtain different types of classified bounding boxes;

[0046] S133, performing semantic parsing on the different types of classification bounding boxes to obtain the non-text content;

[0047] S134: Input the formula content into a preset large language model to extract core symbols in the formula content to obtain key formula information;

[0048] S135, inputting the chart content into the large language model to extract key conclusions from the chart content to obtain key chart information;

[0049] S136: Input the image content into the large language model to extract the function description in the image content to obtain key image information;

[0050] S137: Use the key formula information, the key chart information, and the key image information as the key non-text information.

[0051] Specifically, the slide is scanned using OCR technology or a convolutional neural network to frame the boundary coordinates of the formula / chart / image content; the boundary boxes corresponding to the boundary coordinates are classified according to visual features to obtain different types of classification boundary boxes, wherein the visual features include formulas (including mathematical symbols), data charts (bar / pie charts), and schematic diagrams (flow charts, etc.); semantic analysis is performed on the different types of classification boundary boxes to obtain the non-text content, for example, the formula classification boundary box analysis conforms to the relationship, and the formula content, the chart content, and the image content are input into a preset large language model to obtain key formula information, key chart information, and key image information, wherein the large language model is a deep learning model based on the Transformer architecture. Through self-supervised learning of massive text, it masters the probability distribution law of human language. Its core capabilities include language generation, semantic understanding, and knowledge reasoning. Language generation is to generate coherent text based on context, semantic understanding is to parse the implicit intention of complex instructions; knowledge reasoning is cross-domain logical deduction; in the present invention, the large language model is trained to optimize model parameters to process non-text content and obtain key non-text information. Finally, the key formula information, the key chart information and the key image information are used as the key non-text information.

[0052] S140 , generating a target video using a preset pseudo-autoregressive model according to the narration content corresponding to the side voice and the highlighted slide.

[0053] Specifically, the pseudo-autoregressive model combines the advantages of the autoregressive (AR) and non-autoregressive (NAR) models. Through dynamic length span generation, it maintains the temporal modeling capability of the AR model and utilizes the parallel generation advantage of the NAR model. In the present invention, the pseudo-autoregressive model is used to achieve intelligent synchronization of narration content and highlighted slides. It should be noted that the autoregressive model is generated strictly sequentially, and the output at each moment depends on the results of all previous moments, such as LSTM and Transformer decoders; the non-autoregressive model generates outputs at all moments in parallel, eliminating temporal dependencies, such as BERT (mask prediction).

[0054] Among them, such as Figure 4 As shown, step S140 includes the following steps: S141-S142:

[0055] 141. Inputting the narration content corresponding to the side voice and the highlighted slide into the preset pseudo-autoregressive model to adjust the highlighted slide to obtain a target highlighted slide;

[0056] 142. Generate the target video according to the narration content and the target highlight slide.

[0057] Specifically, the narration content corresponding to the side voice is input into the preset pseudo-autoregressive model to segment the narration content into a phoneme sequence, extract the speech rate feature of the phoneme sequence, and mark the keyword position to obtain an emphasis mark; the highlighted slide is input into the pseudo-autoregressive model to obtain the highlighted area coordinates and the effect parameters within the highlighted area coordinates; the target highlighted slide is generated according to the speech rate feature, the emphasis mark, and the effect parameters. It should be noted that, in this embodiment, the phoneme sequence is, for example, {"phoneme":"grow","start":12.3s,"end":12.8s}; the speech rate feature includes the factor duration and intonation intensity; the keyword position is, for example, the term appearance timestamp; the highlighted area coordinates are the bounding box vector [x1, y1, x2, y2], and the effect parameters include transparency, color mode, and animation type. It should also be noted that in this embodiment, the pseudo-autoregressive model includes a dynamic span predictor, non-autoregressive parallel decoding, and an autoregressive boundary corrector. The dynamic span predictor predicts and processes a parallel span N (N∈[3,8] phonemes) based on speech rate characteristics and emphasis markers. Non-autoregressive parallel decoding then synchronously processes the effect parameters corresponding to the N phonemes. The autoregressive boundary corrector fine-tunes the N highlight boundaries to ensure a smooth transition. Understandably, if the speech rate in the voiceover increases by 15%, the adjustment logic is to compress the ΔT duration, and the target highlight slide output results in a 22% reduction in highlight duration. If continuous terms appear in the voiceover content, the adjustment logic is to adopt a hierarchical transparency approach, and the target highlight slide output results in a layered highlighting of primary and secondary content.

[0058] Through steps S141-S142, the highlighted slides are dynamically adjusted according to the narration content through a pseudo-autoregressive model, so that the narration content is aligned with the adjusted highlighted slides, effectively improving the generated video quality and generation efficiency, and avoiding the low efficiency and error-prone problems of manual video editing.

[0059] The present invention adopts a method for automatically generating slide videos. The method first converts key text information in a slide into narration voice, where the key text information is obtained by fusing multi-source information. Then, the non-text content in the slide is processed to obtain key non-text information, and the key non-text information in the slide is highlighted to obtain a highlighted slide. Finally, the narration content corresponding to the narration and the highlighted slide are input into a preset pseudo-autoregressive model to generate a more natural and smooth target video. During the target video generation process, the highlighted slide is dynamically adjusted according to the narration content through the pseudo-autoregressive model, so that the narration content is aligned with the adjusted highlighted slide, thereby effectively improving the quality and generation efficiency of the generated video.

[0060] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0061] The non-Company software tools or components appearing in the embodiments of this application are merely examples and do not represent actual use.

[0062] In one embodiment, a slide video automatic generation device 200 is provided, which corresponds to the slide video automatic generation method in the above embodiment. Figure 5 As shown, the slide video automatic generation device includes an acquisition unit 201, a processing and conversion unit 202, an extraction and display unit 203, and a generation unit 204. The functional modules are described in detail as follows:

[0063] An acquiring unit 201 is configured to acquire a slide file, wherein the slide file includes a plurality of slides and explanation information corresponding to each slide;

[0064] a processing and conversion unit 202 for processing the text content in the slide according to the explanation information and the input target video description information to obtain key text information, and converting the key text information into narration voice;

[0065] an extraction and display unit 203 for extracting non-text content from the slide using computer vision technology, processing the non-text content to obtain key non-text information, and highlighting the key non-text information in the slide to obtain a highlighted slide;

[0066] The generating unit 204 is configured to generate a target video according to the narration content corresponding to the side voice and the highlighted slides through a preset pseudo autoregressive model.

[0067] In one embodiment, the processing conversion unit 202 is specifically configured to:

[0068] Obtaining the target video description information;

[0069] Inputting the text content in the slide, the target video description information, and the explanation information into a key information extraction model to extract key information from the text content, wherein the key information extraction model is a model based on artificial intelligence and natural language processing technology;

[0070] Marking the position of the key information in the slide to obtain the key text information;

[0071] Detect whether the target timbre information is input;

[0072] If the target timbre information is input, the target timbre information and the key text information are converted into the narration voice by using TTS technology;

[0073] If the target timbre information is not input, default timbre information is obtained, and the default timbre information and the key text information are converted into the narration voice through the TTS technology.

[0074] In one embodiment, the extraction display unit 203 is specifically configured to:

[0075] Scanning the slide using the computer vision technology to determine the boundary coordinates of the non-text content;

[0076] Classifying the bounding boxes corresponding to the bounding coordinates according to visual features to obtain different types of classified bounding boxes;

[0077] performing semantic parsing on the different types of classification bounding boxes to obtain the non-text content;

[0078] Inputting the formula content into a preset large language model to extract core symbols in the formula content to obtain key formula information;

[0079] Inputting the chart content into the large language model to extract key conclusions from the chart content to obtain key chart information;

[0080] Inputting the image content into the large language model to extract the functional description in the image content to obtain key image information;

[0081] The key formula information, the key chart information, and the key image information are used as the key non-text information.

[0082] In one embodiment, the generating unit 204 is specifically configured to:

[0083] Inputting the narration content corresponding to the side voice and the highlighted slide into the preset pseudo-autoregressive model to adjust the highlighted slide to obtain a target highlighted slide;

[0084] The target video is generated according to the narration content and the target highlight slide.

[0085] In one embodiment, the generating unit 204 is further configured to:

[0086] Inputting the narration content corresponding to the side voice into the preset pseudo-autoregressive model to segment the narration content into a phoneme sequence, extracting the speech rate feature of the phoneme sequence, and marking the keyword position to obtain an emphasis mark;

[0087] Inputting the highlighted slide into the pseudo-autoregressive model to obtain the highlighted region coordinates and the effect parameters within the highlighted region coordinates;

[0088] The target highlight slide is generated according to the speech rate feature, the emphasis mark, and the effect parameter.

[0089] The present invention uses a slide video automatic generation device to first convert key text information in a slide into narration voice, where the key text information is obtained by fusing multi-source information; then, the non-text content in the slide is processed to obtain key non-text information, and the key non-text information in the slide is highlighted to obtain a highlighted slide; finally, the narration content corresponding to the narration and the highlighted slide are input into a preset pseudo-autoregressive model to generate a more natural and smooth target video. During the target video generation process, the highlighted slide is dynamically adjusted according to the narration content through the pseudo-autoregressive model, so that the narration content is aligned with the adjusted highlighted slide, thereby effectively improving the generated video quality and generation efficiency.

[0090] The specific definition of the automatic slide video generation device can be found in the definition of the automatic slide video generation method above, and will not be repeated here. The various units in the above-mentioned automatic slide video generation device can be implemented in whole or in part by software, hardware, or a combination thereof. The above-mentioned units can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0091] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 6 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, it realizes the functions or steps on the server side of a method for automatically generating slide videos.

[0092] In one embodiment, a computer device is provided. The computer device may be a client, and its internal structure diagram may be as follows: Figure 7As shown. The computer device includes a processor, memory, network interface, display screen and input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps on the client side of a method for automatically generating a slide video.

[0093] In one embodiment, a computer device is provided, 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, the steps of the above-mentioned method for automatically generating a slideshow video are implemented.

[0094] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned method for automatically generating a slide video are implemented.

[0095] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can be found in the relevant descriptions of the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0096] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchl ink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0097] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0098] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A method for automatically generating a slideshow video, characterized in that: include: Obtaining a slide file, wherein the slide file includes multiple slides and explanation information corresponding to each slide; Processing the text content in the slideshow according to the explanation information and the input target video description information to obtain key text information, and converting the key text information into narration voice; extracting non-text content from the slide using computer vision technology, processing the non-text content to obtain key non-text information, and highlighting the key non-text information in the slide to obtain a highlighted slide; A target video is generated according to the narration content corresponding to the side voice and the highlighted slides through a preset pseudo-autoregressive model.

2. The method for automatically generating a slideshow video according to claim 1, wherein: The step of processing the text content in the slide according to the explanation information and the input target video description information to obtain key text information includes: Obtaining the target video description information; Inputting the text content in the slide, the target video description information, and the explanation information into a key information extraction model to extract key information from the text content, wherein the key information extraction model is a model based on artificial intelligence and natural language processing technology; The key text information is obtained by marking the position of the key information in the slide.

3. The method for automatically generating a slideshow video according to claim 1, wherein: The step of converting the key text information into narration voice includes: Detect whether the target timbre information is input; If the target timbre information is input, the target timbre information and the key text information are converted into the narration voice by using TTS technology; If the target timbre information is not input, default timbre information is obtained, and the default timbre information and the key text information are converted into the narration voice through the TTS technology.

4. The method for automatically generating a slideshow video according to claim 1, wherein: The step of extracting the non-text content in the slide using computer vision technology includes: Scanning the slide using the computer vision technology to determine the boundary coordinates of the non-text content; Classifying the bounding boxes corresponding to the bounding coordinates according to visual features to obtain different types of classified bounding boxes; Semantic parsing is performed on the different types of classification bounding boxes to obtain the non-text content.

5. The method for automatically generating a slideshow video according to claim 1, wherein: The non-text content includes formula content, chart content, and image content. The step of processing the non-text content to obtain key non-text information includes: Inputting the formula content into a preset large language model to extract core symbols in the formula content to obtain key formula information; Inputting the chart content into the large language model to extract key conclusions from the chart content to obtain key chart information; Inputting the image content into the large language model to extract the functional description in the image content to obtain key image information; The key formula information, the key chart information, and the key image information are used as the key non-text information.

6. The method for automatically generating a slideshow video according to any one of claims 1 to 5, wherein: The step of generating a target video according to the narration content corresponding to the side voice and the highlighted slide using a preset pseudo-autoregressive model includes: Inputting the narration content corresponding to the side voice and the highlighted slide into the preset pseudo-autoregressive model to adjust the highlighted slide to obtain a target highlighted slide; The target video is generated according to the narration content and the target highlight slide.

7. The method for automatically generating a slideshow video according to claim 6, wherein: The step of inputting the narration content corresponding to the side voice and the highlighted slide into the preset pseudo-autoregressive model to adjust the highlighted slide to obtain a target highlighted slide includes: Inputting the narration content corresponding to the side voice into the preset pseudo-autoregressive model to segment the narration content into a phoneme sequence, extracting the speech rate feature of the phoneme sequence, and marking the keyword position to obtain an emphasis mark; Inputting the highlighted slide into the pseudo-autoregressive model to obtain the highlighted region coordinates and the effect parameters within the highlighted region coordinates; The target highlight slide is generated according to the speech rate feature, the emphasis mark, and the effect parameter.

8. A device for automatically generating slideshow videos, characterized in that: include: An acquiring unit, configured to acquire a slide file, wherein the slide file includes a plurality of slides and explanation information corresponding to each of the slides; a processing and conversion unit, configured to process the text content in the slide according to the explanation information and the input target video description information to obtain key text information, and convert the key text information into narration voice; an extraction and display unit, configured to extract non-text content from the slide using computer vision technology, process the non-text content to obtain key non-text information, and highlight the key non-text information in the slide to obtain a highlighted slide; A generating unit is used to generate a target video according to the narration content corresponding to the side voice and the highlighted slides through a preset pseudo autoregressive model.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method for automatically generating a slide video according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for automatically generating a slideshow video as claimed in any one of claims 1 to 7 are implemented.

Images