Turning a lecture channel into a searchable Obsidian vault

Published: October 27, 2026 (UTC)

If you study from YouTube (a professor's lecture channel, a course playlist, a series of conference talks), the material is great and hard to search. You remember a professor explained something important, but not which of forty videos it was in, or where in a ninety-minute lecture. This is a workflow for turning a lecture channel into something you can search, inside Obsidian or any Markdown-based notes system.

Step 1: get the course into your library. Run a channel extraction on the lecture channel, or pick the videos you want, and each lecture is transcribed and saved with timestamps. Shorts are included by default, with a checkbox to leave them out. Worth knowing for lecture content: a real share of it has no usable YouTube caption track, such as older recordings, live-captured sessions never post-processed, or non-English talks with poor auto-captions. Channel jobs are captions-only by default. On the plans that include Whisper for jobs, you can switch to "Captions, then Whisper" so uncaptioned lectures get transcribed from the audio too. A single video you paste directly falls back to Whisper on its own, within your monthly Whisper allowance.

Step 2: export to Markdown with clickable timestamps. ChannelScout's Markdown export doesn't just dump text: each timestamp is a real link back to that exact second of the YouTube video, and the file starts with the title, channel and video link. Paste or save that into Obsidian, and you have a note where every claim is one click from its source.

Step 3: drop the notes into your vault. It's plain Markdown, so there is no import plugin to fight with. Save the export as a .md file in your vault like any other note, link it from your course index, tag it, and backlink specific timestamps from your own summary notes.

Step 4: search the whole course, not one lecture at a time. Once several lectures are in your library, ChannelScout's search runs keyword, boolean, proximity or regex queries across all of them, so "which lecture covered X" becomes a five-second search, before you've opened Obsidian. If you want a quick overview of a lecture first, the AI summary gives you a TL;DR, key points and notable quotes.

Step 5: pull the exact clip. If you need to show an explanation rather than describe it, such as a derivation on a whiteboard, the Clip Builder cuts it directly from the transcript view, with no screen-recording software.

Why this matters: AI study tools like NotebookLM read YouTube's caption track, so a video without captions can't be imported. Transcribing first and exporting to Markdown sidesteps that. You hand every downstream tool a transcript it can read, not a YouTube link it has to fail on.