Built this because Chrome's Live Caption is English-only and every "translate any video" tool is paid SaaS. Wanted Arabic subs on English YouTube lectures running on my own GPU.
Stack: MV3 extension captures tab audio → WebSocket → Python backend running faster-whisper + deep-translator → JSON transcripts back → overlay on the video element. Native Messaging host so clicking Start launches the backend itself, no separate terminal.
Real-time on a 4070 (~200ms per chunk). Works on anything that isn't DRM-protected — YouTube, Twitch, podcasts, lectures, free streamers.
Windows-only auto-installer right now. Linux/macOS users can run the backend manually. PRs for cross-platform installers very welcome.
Built this because Chrome's Live Caption is English-only and every "translate any video" tool is paid SaaS. Wanted Arabic subs on English YouTube lectures running on my own GPU.
Stack: MV3 extension captures tab audio → WebSocket → Python backend running faster-whisper + deep-translator → JSON transcripts back → overlay on the video element. Native Messaging host so clicking Start launches the backend itself, no separate terminal.
Real-time on a 4070 (~200ms per chunk). Works on anything that isn't DRM-protected — YouTube, Twitch, podcasts, lectures, free streamers.
Windows-only auto-installer right now. Linux/macOS users can run the backend manually. PRs for cross-platform installers very welcome.
MIT.