SubExtractor

Real-time AI transcription and subtitle extraction for video content

SubExtractor turns any video into clean, searchable text, entirely on your device. Whether you're studying a talk, taking meeting notes, or saving a transcript, SubExtractor transcribes audio and pulls existing captions right in a side panel. No account, no cloud upload, no subscription. Your audio and text never leave your browser. What it does 🎙️ On-device transcription: capture any tab's audio and transcribe it with a local AI speech model. Nothing is sent to a server. 🗣️ Speaker identification (new in 4.0): live transcripts label who's speaking. 📝 Site caption extraction: instantly grab existing subtitles, with clickable timestamps, on supported video sites, plus a general fallback for many others. ⏱️ Timestamped and seekable: click any line to jump straight to that moment in the video. ✨ AI clean-up: fix punctuation, capitalization, and spacing to turn raw captions into readable prose. 🧾 AI summaries: get a TL;DR plus key points, each linked to the moment it's discussed. 🔎 Search, edit, copy and download: full-text search, an editable transcript, and one-click copy or .txt export. Supported sites On-device transcription works on any tab that plays audio. Caption extraction is built site by site, because every video player stores subtitles differently. It covers a growing list of popular video and news sites, and a general fallback handles many others. Some sites aren't supported yet. The current list is on the support page, where you can also request a site. Private by design Transcription, clean-up, and summaries all run locally using on-device AI. There are no servers, no sign-in, and no analytics. The text is yours. Great for Students, researchers, journalists, note-takers, language learners, and anyone who'd rather read than re-watch. Good to know AI clean-up and summaries use Chrome's built-in on-device AI, which is available on supported desktop hardware and recent Chrome versions. Where it isn't available, SubExtractor automatically switches to standard clean-up, and caption extraction and transcription keep working.