Study a long video course from a combined caption file
At a glance
A student combines a YouTube course's captions before asking NotebookLM questions across lectures. This workflow is listed under Learning, Video Courses, Study Notes.
- Agents / tools
- YT Transcriber, YouTube, NotebookLM
SourceBuilding a tool for these workflows? See weekly rates.
What they did
1) Have a captioned, publicly accessible YouTube playlist and a NotebookLM account. The author's browser route uses YT Transcriber at https://13-234-185-132.sslip.io/; Google sign-in and sufficient extraction allowance are needed for a full playlist. 2) Paste the playlist URL into the tool, select the caption language, start extraction, and inspect the per-video statuses for missing captions. 3) Download the combined Markdown export, retaining the lecture titles and order. The author describes splitting large exports between videos at 450,000 words; keep each uploaded part below NotebookLM's source cap. 4) In a notebook, choose Add sources and upload the Markdown parts, renaming them to .txt if the upload dialog requires it. Add slides separately for material that was only shown on screen. 5) Ask for an outline by lecture heading, then generate questions that combine multiple lectures. Check answers against the cited lecture sections.
Worth knowing
This reads existing captions, not audio. Auto-captions can misread technical terms. The tool currently displays an extraction-delay notice; private and captionless videos are unsuitable.
