VideoCaptioner is an open-source, LLM-powered video subtitling assistant that covers the whole captioning workflow in one tool. It has earned more than 16,300 stars on GitHub and was actively updated in 2026, with both a command-line interface and a desktop GUI.
The tool takes a video from raw footage to finished, subtitled output: it transcribes speech, splits and corrects the subtitle lines with a large language model, translates them, and can burn the result into the video or even generate a dubbed audio track. Several speech-recognition and translation options work completely free with zero configuration, while connecting your own LLM API key unlocks AI-based subtitle refinement and higher-quality translation. It installs in one step from PyPI.
The workflow is deliberately one-stop. The same tool that transcribes a video can also download it from YouTube or Bilibili first, then split over-long subtitle lines into readable segments, proofread them with an LLM, translate them while preserving timing, and finally either burn the captions into the picture as hard subtitles, attach them as a soft track, or generate a dubbed voice-over from the translated text. The desktop GUI adds visual proofreading for creators who want to eyeball the result before exporting, while the CLI suits batch and scripted production pipelines.
VideoCaptioner is free and open source under the GPL-3.0 license. Its built-in speech recognition and Bing/Google translation paths cost nothing; LLM-powered features use your own API key, so you only pay your model provider.
Video creators, online-course producers, translators and social-media teams who regularly need accurate, translated subtitles without manual line-by-line work.