Things I’ve built
agent-caption.
Open source · Python
22 stars on GitHub
MIT licensed
Problem solved
Captions for any video or song, in any language, 100% local with no cloud. The audio never leaves your machine.
What I built
Hinglish-first, speech and music, forced-alignment timing so every word lands on the waveform. Free and MIT. It works with any AI agent (Claude Code, Cursor, GitHub Copilot, ChatGPT, Gemini, Grok, or any other): open the folder and say "caption ./myvideo.mp4".
03 / 16