What is video-captions?

video-captions is a set of shell scripts that automatically adds burned-in captions (subtitles permanently embedded into the video picture) to any video file on a Mac. You give it a video, it converts the speech to text using whisper.cpp, a fast, locally-running transcription tool, and then uses ffmpeg to render the captions into the video. The whole process runs entirely on your own computer with no cloud service or internet connection required once set up. There are two ways to use it. The simplest is a single command that does everything at once: transcribe the audio and burn in the captions, writing a new video file next to the original. The two-step path lets you review and edit the auto-generated subtitle file before burning it in, useful when whisper.cpp mishears something. The tool is notably fast. A 6-minute 1080p60 video processes in roughly 2 minutes 40 seconds on an M3 Mac, compared to over 3 hours with a slower approach. Two design choices produce this speed: whisper.cpp uses Apple Silicon's Metal GPU acceleration for transcription, which is much faster than running the same model in Python, and the burn-in step splits the video into parallel segments that are all rendered at the same time and then joined together. The caption style (font, size, color, placement) is fixed in the script and must be edited in code to change. The tool is macOS-only and English-only by default, and requires Homebrew to install its dependencies.