Free Video Subtitle Extractor - OCR to SRT, VTT, TXT | 2026
Extract hardcoded subtitles from any video using browser OCR. Supports Korean, English, Japanese, Chinese, Arabic and 100+ languages. Export as SRT, VTT, or plain text. No upload, no sign-up.
Pulling Hardcoded Subtitles Out of a Video as Text
Overview
When subtitles are burned into a video, the text isn't stored as data you can copy — it's just pixels. A subtitle extractor uses OCR (optical character recognition) to read those on-screen captions frame by frame and turn them back into editable text, exported as SRT, VTT, or plain TXT. It supports Korean, English, Japanese, Chinese, Arabic, and 100+ languages, and runs entirely in your browser with no upload and no sign-up.
How to Use (Step by Step)
- 1
Load the video and mark the caption area
Drop the file and draw a box over the region where subtitles appear, so OCR focuses there.
- 2
Run OCR
The tool reads the captions frame by frame and detects timing. Pick the caption language for best accuracy.
- 3
Export SRT, VTT, or TXT
Download a timed subtitle file (SRT/VTT) or plain text. Review and fix any OCR errors before using it.
How It Works
You mark the region where subtitles appear, and the tool samples frames across the video, runs OCR on that region, and detects when the text changes to determine each caption's start and end time. The recognized text plus timing is assembled into a standard subtitle file. For SRT and VTT, timestamps are generated automatically; TXT gives you just the dialogue. OCR runs locally via WebAssembly, so the video and the extracted text stay on your device.
When to Use This
Recovering a subtitle file from a video that only has burned-in captions. Translating captions later (extract to SRT, then run it through a subtitle translator). Making a transcript of a captioned video. Repurposing dialogue for blog posts or show notes. Re-timing or editing subtitles that were only ever hardcoded. Archiving the text of educational or interview videos.
Frequently Asked Questions
Accuracy is high for clear, high-contrast captions in a supported language, but stylized fonts, low resolution, or busy backgrounds reduce it. Always proofread the output — OCR may misread similar characters. Selecting the correct language improves results.