Full-automatic video generation system and method based on digital twinning and AI technology

Abstract

The invention discloses a full-automatic video generation system and method based on digital twinning and AI technologies, relates to the technical field of image data processing, and aims to solve the problem that full-process automatic production is difficult to realize in existing meteorological and ocean forecast video production. The method comprises the following steps: automatically acquiring structured data from an external digital twin API (Application Program Interface) through an automatic data acquisition module; a broadcast text is automatically generated and synthesized into an audio file based on the structured data through an AI audio automatic generation module; through a dynamic visualization rendering and capturing module, generating a dynamic visualization picture based on structured data driving, capturing the picture to form a continuous image frame sequence, and recording a timestamp of specific element switching in the picture at the same time; and synchronously aligning the voice audio file and the continuous image frame sequence according to the timestamp through an audio and video intelligent packaging module, mixing background music, and packaging and outputting a final video file. And full-process automatic production from structured data to flaking is realized.

Description

Technical Field

[0001] This invention relates to the field of image data processing technology, specifically a fully automated video generation system and method based on digital twin and AI technologies. Background Technology

[0002] The current production of meteorological and marine forecast videos still heavily relies on manual and semi-automated processes, exhibiting significant shortcomings: At the data level, multi-source, heterogeneous forecast data (such as API interfaces and local documents) requires manual extraction and format conversion, resulting in low efficiency and a high risk of errors; at the content production level, professional personnel are required for everything from writing broadcast text and recording professional voice to dynamic visualization design, leading to long production cycles, high costs, and difficulty in meeting the demands of real-time, high-frequency updates; in the compositing stage, audio-visual synchronization largely relies on manual alignment in post-production, making accurate synchronization difficult and impacting the viewing experience; furthermore, existing systems often suffer from poor platform compatibility, limited output formats, and a lack of effective error tolerance and human intervention mechanisms, resulting in insufficient overall robustness and flexibility. Therefore, there is an urgent need for a video generation system that can achieve full automation from data to finished product, high-precision synchronization, multi-format output, and support for human-machine collaboration. Summary of the Invention

[0003] To address the aforementioned issues, namely the problems raised in the background technology, this invention proposes a fully automatic video generation system and method based on digital twin and AI technologies. The system includes an automatic data acquisition module, which is used to acquire structured meteorological or marine forecast data from an external digital twin API at regular intervals through a preset authentication interface. It also includes a built-in fault-tolerant mechanism that automatically retryes at preset intervals after acquisition failure, and provides a manual import interface for local structured data files. The AI ​​audio automatic generation module is used to input the structured meteorological or marine forecast data into a natural language generation engine based on domain knowledge rules to generate broadcast text that conforms to professional broadcasting standards, and then synthesize the broadcast text into corresponding audio files through a speech synthesis model. The dynamic visualization rendering and capture module is used to drive the local visualization rendering engine to generate dynamic images bound to the data in real time based on the structured meteorological or marine forecast data, and to capture frames at equal time intervals during the rendering process of the dynamic images to generate a continuous sequence of image frames. During the rendering process, the timestamps that identify the data content or visual scene switching points are recorded synchronously. The audio and video intelligent encapsulation module is used to obtain the timestamp, synchronize and mix the audio file and the image frame sequence based on the timestamp, and encapsulate them to generate the final video file.

[0004] A further provision of the present invention is that the natural language generation engine is used to perform at least one of the following processing on marine forecast data: converting numerical values ​​and units into colloquial expressions; organizing discrete data entries into logically coherent descriptive sentences; serializing tabular tidal time series data into text broadcast in chronological order; and identifying and adapting text generation or omission processing for default or outlier values ​​in the data.

[0005] A further provision of the present invention is that the speech synthesis model is used to call the local GPT-SoVITS speech synthesis service through a cross-platform adaptation script, and pass the broadcast text and selected speech style parameters to the GPT-SoVITS service to generate a speech audio file of a specific style; the adaptation script includes a .bat batch file for Windows systems and a .sh script for Linux systems.

[0006] A further provision of the present invention is that: the dynamic visualization rendering and capture module is used to dynamically load and display the corresponding background image based on the geographical area information in the structured meteorological or marine forecast data; to render and overlay the numerical information in the structured meteorological or marine forecast data in real time as dynamically changing numbers, charts, or animated icons at a specified position on the background image, thereby forming a dynamic picture containing a background image and a dynamic data layer; and to capture frames at equal time intervals during the rendering process of the dynamic picture at a fixed frame rate to generate a continuous image frame sequence.

[0007] A further provision of the present invention is that: the audio and video intelligent encapsulation module uses the audio and video processing tool FFmpeg to apply corresponding delays to multiple audio files based on the timestamp, so as to achieve millisecond-level synchronization with the scene switching; the synchronized multiple audio files, background music and opening music are mixed to generate a single audio track; the single audio track and the image frame sequence are encoded and encapsulated according to a preset video encoding standard.

[0008] A further provision of the present invention is that the automatic data acquisition module supports parsing locally submitted formatted documents, the format of which includes .doc, .docx, and .xml files; or it can call an external API through a network interface to obtain real-time data in JSON or XML data exchange format.

[0009] A further provision of the present invention is that the video encoding standard supports at least the generation of a first format file for television broadcasting and a second format file for network transmission, wherein the first format file is an MXF format file based on MPEG-2 encoding, the second format file is an MP4 format file based on H.264 encoding, and the system supports the generation of video files with 4K or 1080P resolution.

[0010] A further provision of the present invention is that the system also includes a user intervention module, which provides at least one of the following manual intervention functions: manually editing the data imported or acquired by the automatic data acquisition module; changing the static background image used by the dynamic visualization rendering and capture module; adjusting the volume of the voice audio file generated by the AI ​​audio automatic generation module or selecting different voice styles; adjusting the volume of the background music or selecting different background music; previewing the synthesis effect; and selecting the resolution and format of the final output video.

[0011] A fully automated video generation method based on digital twin and AI technologies includes: The automatic data acquisition module periodically retrieves structured data from external digital twin APIs. The AI ​​audio automatic generation module automatically generates broadcast text based on the structured data and synthesizes it into a speech audio file. The dynamic visualization rendering and capture module generates dynamic visualization images based on the structured data, captures the images to form a continuous sequence of image frames, and records the timestamps of specific element switching in the images. The audio-video intelligent encapsulation module synchronizes and aligns the audio file with the continuous image frame sequence based on the timestamp, mixes background music, and encapsulates and outputs the final video file. The beneficial technical effects of this invention are as follows: This invention achieves fully automated production from structured data to finished product, significantly improving the efficiency, consistency, and professionalism of video production. The system automatically acquires and intelligently parses multi-source data, supporting seamless integration with APIs and local documents; based on domain-rule-based natural language generation and speech synthesis technology, it can output audio content that conforms to broadcasting standards; combined with data-driven dynamic visualization rendering and a millisecond-level timestamp synchronization mechanism, it achieves precise alignment of audio, video, and background music, meeting broadcast standards. Simultaneously, the system has cross-platform operation capabilities, supporting the generation of multiple versions of videos suitable for television broadcast (such as MXF format) and online distribution (such as MP4 format), and provides full-process visualized intervention nodes, ensuring efficient automated operation while also addressing flexibility needs such as data verification, style adjustment, and output customization. Furthermore, the modular design and script-based calling architecture make the system easy to maintain and expand, exhibiting good engineering applicability and technological foresight. Attached Figure Description

[0012] Figure 1 A simplified flowchart of the video generation method of the present invention is shown.

[0013] Figure 2 A schematic diagram of the video generation system of the present invention is shown. Detailed Implementation

[0014] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0015] This embodiment uses the generation of a "Liaoning Province Marine Weather Forecast" video as an example to illustrate the system's workflow.

[0016] I. The working process of the automatic data acquisition module: The system is configured to run automatically at 8:50 AM daily. First, the automatic data acquisition module securely connects to an external digital twin API via a pre-defined authentication interface (such as a token mechanism) to obtain the latest structured forecast data (JSON format). If the initial request fails due to network fluctuations, the module will activate a fault-tolerance mechanism, automatically retrying every 5 minutes until successful. As a backup, users can also manually upload a compliant local XML data file through the system interface. The system will parse this file to obtain the required structured data. For example, the acquired data might contain the following key fields: {“region”:“dandong”,“wave_height”:“0.7”,“water_temp”:“0.6”,“tide_high_time”:“09:53”,“tide_high”:“524”…}, indicating that the data corresponds to the region Dandong, with a wave height of 0.7 meters, a water temperature of 0.6 degrees Celsius, a high tide time of 09:53, and a high tide level of 524 centimeters…

[0017] II. The working process of the AI ​​audio automatic generation module: The AI ​​audio auto-generation module receives the aforementioned JSON data. Its built-in natural language generation engine, based on knowledge rules for the ocean forecasting field, begins operation.

[0018] 1. Data Extraction, Classification, and Script Synthesis: Accurately identify geographical regions such as "Dandong" and "Dalian," along with corresponding multi-dimensional data including wave height, water temperature, tidal time, and tide height. Data is converted and organized according to a specialized vocabulary and grammatical rules for marine forecasting. For example, the numerical value "0.7" and the unit "meter" are colloquially translated as "0.7 meters"; discrete data lines {"region":"dandong","wave_height":"0.7","water_temp":"0.6"} are organized into a coherent sentence conforming to broadcasting conventions: "In the waters near Dandong, wave height is 0.7 meters, surface water temperature is 0.6 degrees Celsius."; tidal data {"tide_high_time":"09:53","tide_high":"524"} are serialized into: "Dandong Port, the first high tide occurs at 9:53 AM, tide height is 524 centimeters."

[0019] All generated sentences are arranged according to the logical structure of "Summary → Regional Details → Special Forecast (Tides)" and standard introductions such as "Liaoning Provincial Marine Early Warning and Monitoring Center issued..." are added to generate a complete broadcast text suitable for speech synthesis.

[0020] 2. Calling the GPT-SoVITS speech synthesis service (cross-platform script adaptation): The generated broadcast text and the style parameters of "male or female news anchor" selected by the user on the front end are transmitted and synthesized through a cross-platform decoupled call chain. The core design of this call chain is to use a platform-adaptive script as a call intermediary to shield operating system differences and decouple business logic from the underlying speech synthesis service.

[0021] The specific invocation process is as follows: The application does not directly call the core code of the GPT-SoVITS service, but instead executes the corresponding platform script. In a Windows environment, the system calls the pre-configured go-interence-webui.bat batch script; in a Linux environment, it calls the pre-configured go-interence-webui.sh shell script. During the invocation, the application passes the broadcast text, style identifier, and output path as command-line arguments.

[0022] Service calls are triggered by these platform scripts: these scripts encapsulate platform-related details such as starting the Python interpreter, locating the service directory, and setting environment variables. They launch a specified Python client script, which is responsible for constructing the passed parameters into a request format (such as a function call) recognized by the GPT-SoVITS service API. Ultimately, this drives the locally deployed GPT-SoVITS service to complete the speech synthesis task, generating high-quality regional speech files (such as dandong.wav, dalian.wav) and a welcome message file (welcome.wav) in the corresponding style.

[0023] The advantage of this decoupled architecture is that when the internal API or deployment path of the GPT-SoVITS service changes, only the corresponding platform scripts or Python client scripts need to be adjusted, without modifying the core code of the main application, greatly enhancing the system's maintainability and backward compatibility. Through this standardized parameter and scripted call, the system provides a consistent speech synthesis service interface on both Windows and Linux systems, achieving efficient, flexible, and stable collaboration between the broadcast content, user style selection, and the underlying complex TTS services.

[0024] III. The working process of the dynamic visualization rendering and capture module: The dynamic visualization rendering and capture module starts the local visualization rendering engine based on the same JSON data. The specific implementation process is as follows: The application incorporates a visualization engine and UI component library to construct an interface composed of switchable static background layers (such as landmark buildings corresponding to different cities) and dynamic data overlays. When the forecast process begins, the system dynamically loads and switches the display of the corresponding static background image of landmark buildings in the "Dandong" region from the material library based on the geographical area involved in the forecast data (such as the "region" field value being "dandong" in the data). At the same time, various marine data (such as wave height, water temperature, and other numerical information) are dynamically bound to and refreshed in real time on the designated preset positions of the UI through dynamic scrolling numbers, animated waveforms, or charts, forming a comprehensive information map that dynamically updates with the forecast content. When the data is processed to the next region (such as "Dalian"), the background image seamlessly switches, and the visualized data is updated in real time.

[0025] The system controls the entire dynamic rendering process to run precisely at a fixed frame rate (25 frames / second) for a preset duration (e.g., 60 seconds). During this process, location transitions, data item transitions, and numerical changes are all presented in a smooth animation. Throughout the animation, the system simultaneously executes two key tasks: 1. Frame capture: Captures the application window at equal time intervals, generating a series of continuous static image frames, and saves them to the specified directory in sequence (e.g., frame_0000.jpg, frame_0001.jpg, ..., frame_1499.jpg), capturing a total of 1500 frames to correspond to a 60-second video duration.

[0026] 2. Timestamp Recording: The system will accurately record the time when each key action occurs (timestamp), such as the start time when the background switches to the "Dandong" screen (dandong_start_time=12000ms, i.e., the 12th second), and the time when the screen switches to the "Dalian" screen, etc.

[0027] This module transforms structured data into visual animations with geographic features and dynamic effects. Through automated, high-fidelity frame sequence capture and precise marking of key moments, it provides subsequent audio-visual synthesis with video stream material that strictly matches the duration of the broadcast content and whose visual content changes synchronously with the forecast logic. It also provides a time reference for precise alignment, thus achieving seamless synchronization of data, audio, and video. Finally, the generated continuous image frame sequence (e.g., frame_0001.jpg to frame_1500.jpg) and the recorded timestamp list serve as the output of this module.

[0028] IV. Working process of the audio and video intelligent encapsulation module: This module is the final synthesis stage of the system. Its core function is to use the precise timestamps obtained from the dynamic rendering process to synchronously mix multiple independently generated, geographically segmented audio segments, background music, and title music with a continuous sequence of video frames at the millisecond level, and encapsulate them into video files that meet broadcast-grade or general playback standards, thereby achieving precise alignment between audio content and visual scene changes.

[0029] The specific implementation process and technical details are as follows: Step 1: Preparation of Compositing Parameters and Gathering of Input Materials: The system gathers all intermediate materials and synchronization parameters generated by the aforementioned modules, specifically including: 1. Video stream material: A continuous sequence of image frames from the dynamic visualization rendering and capture module (e.g., 1500 JPG images, corresponding to a duration of 60 seconds, with a frame rate of 25 frames per second).

[0030] 2. Audio stream materials: Audio broadcasts from different regions from the AI ​​audio automatic generation module (such as welcome.wav, dandong.wav, dalian.wav).

[0031] 3. Pre-set background music files (such as bg.mp3).

[0032] 4. Pre-set intro / title music files (such as title.mp3).

[0033] 5. Key synchronization parameters: timestamps (in milliseconds) recorded during dynamic rendering. Each timestamp marks the precise start time when the scene switches to a specific region (such as Dandong or Dalian) or plays a specific element.

[0034] Step Two: Timestamp-Based Multitrack Audio Alignment and Mixing: The system utilizes FFmpeg's filter functionality to synthesize and synchronize audio based on timestamps. An adelay filter is applied to each regional audio WAV file, with the delay time strictly corresponding to the scene transition timestamp for that region. For example, the delay of dandong.wav is set to dandong_start_time (12000 milliseconds). This ensures that when the video screen switches to the "Dandong" background, the corresponding "Dandong Sea Area Forecast" audio begins playing synchronously.

[0035] Independent volume gain control is applied to all audio streams, including aligned speech, background music, and intro / title music. Background music is trimmed and faded out to ensure a smooth ending at the end of the video. Finally, an amix filter is used to blend all audio streams into a single, fixed-duration final audio stream (mixed_audio.wav).

[0036] Step 3: Cross-platform standardized video encoding and encapsulation. FFmpeg is invoked through a unified command-line interface to perform the final audio and video encapsulation. Video encoding: Input the image frame sequence at a fixed frame rate (e.g., 25fps) and select the encoding standard according to the user-selected configuration.

[0037] The system supports generating videos of different uses and qualities: 1. Broadcast-grade output: Employs MPEG-2 video encoding and PCM audio encoding, encapsulated in MXF format, and follows the BT.709 color space standard to meet professional broadcasting requirements. The generated file is output_broadcast.mxf.

[0038] 2. Universal playback output: It adopts H.264 video encoding and AAC audio encoding, encapsulates it into MP4 format, and optimizes it for web pages (-movflags +faststart) to facilitate network distribution and computer playback, generating files such as output_web.mp4.

[0039] 3. Resolution Adaptive: Automatically sets the output video resolution (e.g., 1920x1080 or 3840x2160) based on user selection (e.g., video_type).

[0040] 4. Cross-platform execution: The main application triggers the compositing process by calling the platform-specific wrapper script (gen_audio.bat on Windows and gen_audio.sh on Linux) and passing in all prepared parameters. This design ensures the consistency and maintainability of the compositing logic across both systems.

[0041] This module solves the problem of lip-syncing and scene synchronization between multi-segment voice broadcasting and dynamically changing visuals by introducing and utilizing a precise timestamp synchronization mechanism. It achieves a fully automated, high-precision production process from raw data to the final product. The output video is not only accurate in content but also achieves professional-level synchronization in terms of audiovisual experience. The generated video files can be directly used for television broadcasting or new media distribution.

[0042] V. User Intervention Module Working Process: This system adopts the design principle of "automatic as the primary method and manual as a secondary method," pre-setting user intervention functions at key nodes. This ensures high-efficiency fully automated production while providing flexibility to handle personalized needs and unexpected situations. Specific intervention nodes and operation methods are integrated into the following core stages: 1. Data import and preprocessing stage: Automation mainline: The system automatically identifies, reads, and parses raw data files obtained from external APIs or manually imported by users.

[0043] User intervention nodes and operations: After completing the initial analysis, the system presents the data in a structured table format on the editing interface. Users can perform manual verification and fine editing here, such as directly clicking on a cell to modify a water temperature value automatically parsed, or deleting or adding entries.

[0044] 2. Content Generation and Configuration Stage Automation Mainline: Based on established rules, the system automatically selects initial visual materials for the project, generates broadcast audio, and configures relevant parameters.

[0045] User Intervention Node and Operation 1 (Visual Material Replacement): In the project editing interface, the system will display automatically matched regional representative images (such as the "Dandong" background image). If the user needs to replace it, they can click the "Replace" button next to the image and upload a new image from the local file system.

[0046] User Intervention Node and Operation 2 (Audio Material Adjustment): AI Dubbing: Users can listen to the pre-generated dubbing generated by the system and select other options from the voice style list provided on the front end (e.g., switch from "Male News Anchor" to "Female News Anchor"), or adjust the dubbing volume individually through the volume adjustment controls.

[0047] Background music: Users can click to select the current background music, or choose other music from the system's recommended track list, and can also adjust its volume independently.

[0048] 3. Preview and Confirmation Stage Automation mainline: The system automatically synthesizes a complete preview version based on all current configurations.

[0049] User Intervention Nodes and Operations: Before final packaging and export, users can click the "Preview" button at any time. The system will render and play a temporary preview video based on all current settings (including any adjustments made by the user). In the preview interface, users can check frame by frame via the timeline or quickly locate playback to verify audio-visual synchronization and content accuracy. If any problems are found, users can return to the aforementioned steps for adjustments.

[0050] 4. Output and Export Stage Automation mainline: The system has built-in default export encoding, format, and resolution settings.

[0051] User intervention nodes and operations: Before exporting, users can flexibly select from preset output options through interface elements such as drop-down menus and button groups. For example, they can make a final decision on combinations of different qualities, formats, and resolutions, such as "4K MXF (TV broadcast)" and "1080P MP4 (online distribution)".

[0052] 5. Global security control User Intervention Nodes and Operations: To prevent accidental manipulation of confirmed parameters during complex editing, the system has placed a "lock" icon next to key adjustable elements (such as confirmed background music volume and voiceover) in the project editing interface. Clicking this icon will lock the corresponding element, making it temporarily uneditable until it is unlocked by clicking it again.

[0053] Through the above-mentioned specific and operable intervention designs that run through the entire process, this system ensures unattended and efficient operation under routine tasks, while providing full control over scenarios such as data accuracy verification, visual style personalization, audio effect optimization, and output format adaptation, significantly improving the system's practicality, adaptability, and user-friendliness.

[0054] Although the invention has been described with reference to preferred embodiments, various modifications can be made thereto and components can be replaced with equivalents without departing from the scope of the invention. In particular, the technical features mentioned in the various embodiments can be combined in any manner, provided there is no structural conflict. The invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

[0055] The term "comprising" or any other similar term is intended to cover non-exclusive inclusion, such that a process, article, or apparatus / device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to those processes, articles, or apparatus / devices.

[0056] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A fully automated video generation system based on digital twin and AI technologies, characterized in that, include: The automatic data acquisition module is used to periodically acquire structured meteorological or marine forecast data from an external digital twin API through a preset authentication interface. It also has a built-in fault tolerance mechanism that automatically retryes at preset intervals after acquisition failure, and provides a manual import interface for local structured data files. The AI ​​audio automatic generation module is used to input the structured meteorological or marine forecast data into a natural language generation engine based on domain knowledge rules to generate broadcast text that conforms to professional broadcasting standards, and then synthesize the broadcast text into corresponding audio files through a speech synthesis model. The dynamic visualization rendering and capture module is used to drive the local visualization rendering engine to generate dynamic images bound to the data in real time based on the structured meteorological or marine forecast data, and to capture frames at equal time intervals during the rendering process of the dynamic images to generate a continuous sequence of image frames. During the rendering process, the timestamps that identify the data content or visual scene switching points are recorded synchronously. The audio and video intelligent encapsulation module is used to obtain the timestamp, synchronize and mix the audio file and the image frame sequence based on the timestamp, and encapsulate them to generate the final video file.

2. The fully automated video generation system based on digital twin and AI technology according to claim 1, characterized in that, The natural language generation engine is used to process marine forecast data in at least one of the following ways: converting numerical values ​​and units into colloquial expressions; organizing discrete data entries into logically coherent descriptive sentences; serializing tabular tidal time series data into text broadcast in chronological order; and identifying and adapting text generation or omission processing for missing or outlier values ​​in the data.

3. The fully automated video generation system based on digital twin and AI technology according to claim 1, characterized in that, The speech synthesis model is used to call the local GPT-SoVITS speech synthesis service through a cross-platform adaptation script, and pass the broadcast text and selected speech style parameters to the GPT-SoVITS service to generate a speech audio file with a specific style; the adaptation script includes a .bat batch file for Windows system and a .sh script for Linux system.

4. The fully automated video generation system based on digital twin and AI technology according to claim 1, characterized in that, The dynamic visualization rendering and capture module is used to dynamically load and display the corresponding background image based on the geographical area information in the structured meteorological or marine forecast data; The numerical information in the structured meteorological or marine forecast data is rendered in real time as dynamically changing numbers, charts, or animated icons and superimposed on a designated position of the background image, thereby forming a dynamic picture containing a background image and a dynamic data layer; the rendering process of the dynamic picture is performed at fixed frame rates with equal time intervals to generate a continuous image frame sequence.

5. The fully automated video generation system based on digital twin and AI technology according to claim 4, characterized in that, The audio and video intelligent encapsulation module uses the audio and video processing tool FFmpeg to apply corresponding delays to multiple audio files based on the timestamp, so as to achieve millisecond-level synchronization with the scene switching. The synchronized audio segments, background music, and opening music are mixed to generate a single audio track; the single audio track and the image frame sequence are then encoded and encapsulated according to a preset video encoding standard.

6. The fully automated video generation system based on digital twin and AI technology according to claim 1, characterized in that, The automatic data acquisition module supports parsing locally submitted formatted documents, including .doc, .docx, and .xml files; or it can call external APIs via network interfaces to obtain real-time data in JSON or XML data exchange formats.

7. The fully automated video generation system based on digital twin and AI technology according to claim 1, characterized in that, The video encoding standard supports at least the generation of a first format file for television broadcasting and a second format file for network transmission. The first format file is an MXF format file based on MPEG-2 encoding, and the second format file is an MP4 format file based on H.264 encoding. The system also supports the generation of video files with 4K or 1080P resolution.

8. A fully automated video generation system based on digital twin and AI technology according to any one of claims 1 to 7, characterized in that, The system also includes a user intervention module, which provides at least one of the following manual intervention functions: manually editing the data imported or acquired by the automatic data acquisition module; changing the static background image used by the dynamic visualization rendering and capture module; adjusting the volume of the voice audio file generated by the AI ​​audio automatic generation module or selecting different voice styles; adjusting the volume of the background music or selecting different background music; previewing the synthesis effect; and selecting the resolution and format of the final output video.

9. A fully automated video generation method based on digital twin and AI technology, characterized in that, The method includes: The automatic data acquisition module periodically retrieves structured data from external digital twin APIs. The AI ​​audio automatic generation module automatically generates broadcast text based on the structured data and synthesizes it into a speech audio file. The dynamic visualization rendering and capture module generates dynamic visualization images based on the structured data, captures the images to form a continuous sequence of image frames, and records the timestamps of specific element switching in the images. The audio-video intelligent encapsulation module synchronizes and aligns the voice audio file with the continuous image frame sequence based on the timestamp, mixes background music, and encapsulates and outputs the final video file.

Images