Batch Video Subtitle Removal Tool

Batch remove hardcoded subtitles from videos with an AI model, intelligently restore the image, and support GPU-accelerated processing.

Bug report

Loading tool, please wait...

If the online tool fails to load or run, try the desktop tool.https://tools.yikeaigc.com/

Tool Usage

Return to old version

A desktop version is available with more powerful, stable features. Open desktop version

Instructions

Software Usage Instructions

Select input files: Click "Choose File" to select one or more video files, or click "Select Folder" to choose a folder containing videos. You can also drag video files directly into the input box. Common video formats such as MP4, AVI, MKV, MOV, WMV, and FLV are supported.
Set output directory: Click the "Browse" button to choose where to save the processed videos. Check "Keep original path structure" to preserve the original folder hierarchy in the output directory.
Adjust subtitle area:
- Top/Bottom: Set the vertical range of the subtitle area (percentage). For example, bottom subtitles are usually set to Top 75% and Bottom 100%.
- Left/Right: Set the horizontal range of the subtitle area (percentage). Usually set Left 5% and Right 95%.
Select detection mode:
- Area coverage (recommended): Directly repairs all content within the specified area, suitable for videos with fixed subtitle positions.
- Threshold detection: Detects subtitle text based on brightness thresholds, suitable for white subtitles.
- Adaptive detection: Automatically adapts subtitle detection to different backgrounds, suitable for complex scenes.
Configure processing options:
- Number of processes: For batch processing, you can enable multi-process parallel processing. Recommended setting: 2–4.
- GPU acceleration: Check this option when you have a dedicated graphics card to greatly improve processing speed.
- Frame skipping acceleration: The higher the value, the faster the speed, but slight flickering may occur.
Preview effect: Click the "Preview Effect" button to view the subtitle removal effect with the current settings and confirm that the area is set correctly.
Start processing: After confirming that the settings are correct, click the "Start Processing" button, and the software will automatically batch-process all video files.

FAQ

A: Area coverage mode repairs all content within the specified area, suitable for videos with fixed subtitle positions and simple backgrounds; Threshold detection mode detects white subtitle text based on brightness, suitable for white subtitles; Adaptive detection mode can automatically adapt to different backgrounds, suitable for videos with varying subtitle colors. Area coverage mode is generally recommended.

A: The subtitle area is defined by four percentage values. Top and Bottom control the vertical range, while Left and Right control the horizontal range. For example, bottom subtitles are usually set to: Top 75%, Bottom 100%, Left 5%, Right 95%. It is recommended to use the "Preview Effect" feature first to confirm that the area is set correctly before starting processing.

A: You can improve processing speed in the following ways: 1) If you have a dedicated graphics card, check the "GPU acceleration" option; 2) Increase the "Number of processes" setting. It is recommended to set it to half the number of CPU cores; 3) Appropriately increase the "Frame skip acceleration" value; 4) Close other resource-intensive programs.

A: No. All video processing is completed locally on your computer. Video files will not be uploaded to any server, ensuring your privacy and data security. The software uses a local AI model to repair subtitle areas.

A: Mainstream video formats such as MP4, AVI, MKV, MOV, WMV, FLV, WebM, M4V, MPEG, MPG, 3GP, and TS are supported. The output format is uniformly MP4, offering good compatibility and making it easy to use later.

A: You can try the following methods to improve the result: 1) Adjust the subtitle area range to ensure it fully covers the subtitles without being too large; 2) Switch detection modes and try the results of different modes; 3) For videos with complex backgrounds, the adaptive detection mode may work better; 4) Make sure the LAMA model file is complete. AI repair performs better than traditional algorithms.

A: The LAMA model file (lama_fp32.onnx) needs to be placed in the same directory as the software. If this file is missing, the software will use the traditional OpenCV inpainting algorithm, and the results may be somewhat reduced. Please make sure to download the complete software package from official channels, including all necessary model files.