Viktor.comAn AI coworker that actually does the work
Promoted
YouTube Chapter Generator
AI chapter detection with visual scene verification
7 followers
AI chapter detection with visual scene verification
7 followers
Unlike tools that only read subtitles, this watches the video. Dual text+visual verification cross-checks topic boundaries against real scene changes for accurate timestamps. Features: upload video/SRT/YouTube link, editable titles, one-click copy to YouTube. Free, offline, zero API. Built with Whisper + CLIP AI.
It would be great if you could add batch processing so I can drop in a whole playlist of YouTube links and get chapters generated for all of them at once, instead of running them one by one.
@ravzapyjn Great suggestion! Batch processing is actually high on our roadmap. Playlist support via yt-dlp is already in the engine — we just need to wire up the UI. Would something like "paste playlist link → generates chapters for all videos at once" cover your use case? Planning to ship this in the next update.
@ravzapyjn Thanks Hanım! Really glad the scene-cut alignment worked on your 40min podcast — that's exactly what the dual verification was designed for. Batch processing is now live too (playlist support + multi-URL). Free on GitHub, Pro with batch on Gumroad. Appreciate you testing it! 🙌
https://github.com/biuta666/youtube-chapter-generator
Would love to see support for translating the chapter titles into other languages since most of my audience watches with auto-translated captions. Could you add a language selector that outputs bilingual timestamps in the YouTube description format?
@erengldalyzxw Love this idea! The engine already generates chapter titles from the transcript text, so swapping the output language is feasible. A bilingual timestamp format (e.g. "03:15 标题 / English Title") could work well for mixed-language audiences. I'll add a language selector to the next release — what languages would be most useful for your audience?
finally tried this on a 40 min podcast and the chapters actually line up with the scene cuts, not random mid-sentence breaks. offline + free is a rare combo for this kind of tool
@hakanaknto2q Thanks Hakan! Really glad the scene-cut alignment worked on your 40min podcast — that's exactly the scenario we built it for. The dual verification (subtitle gaps + visual scene changes) is what keeps those chapter boundaries clean. And yes, keeping it free & offline was non-negotiable for us — no API bills, no data leaving your machine. 🙌
Finally tried this on a 40 min podcast and the chapters lined up way better than the auto ones YouTube gave me, especially around topic shifts that had no subtitle cues. Nice that it runs offline.
@muzdil61976 Thanks Hanım! Appreciate you actually testing it on a real 40min podcast. The topic shifts without subtitle cues are exactly why we added visual scene detection — glad it worked as intended. More improvements coming (batch processing is in progress).