Ghostcut

Ghostcut uses OCR to find text on screen and then paints over it, rebuilding the area underneath so the removal blends into the footage. It handles hard subtitles, captions, lower thirds, watermarks, and logos, and supports Chinese, English, Japanese, Korean, and Arabic text. You can also mark specific regions to remove or protect before the engine runs, which helps when a watermark sits next to something you need to keep.

The platform reads both audio and picture. Automatic speech recognition pulls lines from the spoken track while optical character recognition catches text already on screen, and the two are matched into a timed subtitle file. It supports 100+ languages according to JollyToday, with noise and interference reduction built in. Once the lines are extracted, you can edit them online and tag which speaker said what.

Translation runs through large models such as DeepSeek, and JollyToday describes the flow as multi-agent, meaning one pass drafts and another checks. That setup targets the place where raw machine translation usually falls down: character names, tone, and dialogue that only makes sense in the original language. It's still worth a human read before you publish, because the review step smooths accuracy rather than guaranteeing it.

Ghostcut generates dubbing in dozens of languages, covering US, European, and Asian voices, and supports cloning a speaker's voice from a sample. The goal is a dub that keeps the original performance instead of a flat read. Quality varies by language, and less common languages are where you'll hear the seams most clearly.

Projects and assets are managed in one place, and the platform is built to upload and translate hundreds of videos at once. For teams working through a season or a campaign, that changes the job from a per-file task into a queue you set up once and monitor.

After translation, you composite subtitles, audio, and music back onto the video with automatic alignment, then render the final file. Editing project files can be exported too, so you're not locked into the platform if you want to finish in another editor.

Beyond the web app, Ghostcut offers an API for subtitle removal, aimed at developers and teams who want to wire the capability into their own pipeline. That's the path for anyone pushing volume high enough that manual uploads stop making sense.