YouTube to notes: study notes from a lecture playlist, with a reference per section

You are working through a course on YouTube: twenty lectures, an hour each. By the exam you will remember that lecture seven was good and very little of what was in it. Notes help, but notes you cannot trace back to the lecture are hard to trust when you revise, and impossible to cite.

INDXR turns the playlist into a collection of transcripts, each kept with its source and a ready APA citation. From there you have two ways to get notes: a chapter summary of each video, with a timestamp per chapter, or Claude or ChatGPT writing notes across the whole collection with a reference under every section. Either way, every point leads back to the moment it was said.

A free account includes 10 credits. Pulling YouTube captions is free for a single video, and in a playlist the first 3 caption videos are free, then 1 credit per video.

Step 1: the playlist becomes a collection

Paste the playlist URL into the Playlist tab. INDXR lists every video with its length, you choose which ones to include, and the job runs in the background, up to 500 videos per job. The transcripts are filed into a collection named after the playlist, so the course stays together. Videos without captions can be transcribed with AI at 1 credit per minute. The full flow is in playlist transcripts.

Each transcript keeps its source: the channel, the link, the length, the date it was published, and a citation in APA 7. That is the part that makes notes citable later. How that citation is built is covered in how to cite a YouTube video in APA.

Step 2a: notes across lectures, from Claude or ChatGPT

Connect your library to Claude or ChatGPT once. The connection is read-only and free. The assistant can then list your collections, search the transcripts, read a lecture in blocks with timestamps, and look up each source with its citation. Ask it for notes the way you want them:

“Turn lectures 6 to 10 of my Early Middle Ages collection into study notes. One section per lecture, the key arguments as bullets, and under each section the APA reference and the timestamps you used.”

Because the assistant reads your transcripts rather than recalling the course from memory, it can quote the lecturer and point to the minute. Check a few of those points against the video before you rely on them: the assistant can still misread a passage, and the timestamp is how you find out quickly.

Step 2b: a chapter summary of each video, inside INDXR

For one video at a time, INDXR writes the notes itself. It reads the whole transcript, splits it into chapters where the subject changes, and writes worked-out notes under each one, with a timestamp that jumps the player to that moment.

These are study notes, not an abstract. A fifty-five minute Harvard lecture with a seven thousand word transcript produced around thirty-six hundred words of notes across five chapters. The chapter breaks follow the talk, not a stopwatch. Research on instructional video puts the useful segment at around six to nine minutes, and that is the density we aim for. Where one argument runs longer, it stays in a single chapter rather than being cut in half, because a break in the wrong place costs more than a long chapter does.

Timestamps you can check

A summary you cannot check is a summary you have to trust. Every chapter carries the moment it starts, and clicking it takes the player there. Read the chapter list, find the part that matters, and watch three minutes instead of an hour.

Then make it yours

Open the edit tab and the summary becomes a document you can work in: cut the chapters you do not need, add the connection you saw that the model did not, put your own question at the top. Your version is stored next to the generated one, never on top of it.

What it cannot see

The summary works from the transcript, so it knows nothing about what happens on screen. Slides, diagrams and demonstrations only reach it through what was said about them.

What lands in your notes app

This is a real Markdown export of that Harvard lecture:

---
title: "Justice: What's The Right Thing To Do? Episode 01"
url: "https://www.youtube.com/watch?v=kBdfcR-8hEY"
channel: "Harvard University"
duration: 3282
language: "en"
transcript_source: "YouTube captions"
created: "2026-08-25"
type: youtube
tags: [youtube, summary]
---

# Justice: What's The Right Thing To Do? Episode 01

## Overview

This video introduces the field of moral philosophy by presenting a series of challenging ethical dilemmas, such as the classic trolley problem and variations involving doctors and organ transplants. These scenarios serve to illustrate two fundamental approaches to moral reasoning: consequentialist ethics, which judges actions based on their outcomes, and categorical ethics, which emphasizes inherent duties and rights irrespective of consequences...

## [00:00:00](https://youtu.be/kBdfcR-8hEY?t=0) Introduction to Moral Dilemmas and Reasoning

The course begins with a story about **Justice**.

**The Trolley Problem (Variant 1: Driver)**
*   **Scenario:** You are the driver of a trolley car traveling at 60 mph. Your brakes fail. Ahead are five workers on the track. You will kill all five if you continue.
*   **Alternative:** There is a side track to the right with one worker. Your steering works, allowing you to turn onto the side track, killing the one worker but sparing the five.
*   **Dilemma:** What is the right thing to do? Turn, killing one to save five, or go straight, killing five?

## [00:24:44](https://youtu.be/kBdfcR-8hEY?t=1484) Revisiting Moral Frameworks and Introducing Utilitarianism
...

The front matter is filled from what the video carries: title, channel, duration, language and source. Fields without a value are left out rather than filled with placeholders.

In Obsidian

The front matter maps onto Obsidian Properties without any configuration, so a vault of these notes is queryable the moment you drop them in. Duration is stored in seconds, which makes length comparisons trivial:

TABLE channel, duration / 60 AS "Minutes"
FROM #youtube
WHERE type = "youtube"
SORT duration DESC

In Notion and elsewhere

Notion imports Markdown files directly and turns the headings into a page outline, so the chapter structure survives. It does not map YAML front matter onto database properties automatically, so if you keep a video database there you either paste the fields in once or strip the front matter before importing. Anything else that reads Markdown works too: the file is plain text.

What it costs

Summary cost by video length:

The Harvard lecture above: the transcript cost nothing because the video had captions, and the summary cost 6 credits. No subscription, and credits never expire.

Try it on a course

A free account includes 10 credits, enough for 10 minutes of AI transcription or a few caption-based summaries, with no card. Start with a short playlist, connect Claude or ChatGPT, and ask for notes on one lecture before you do the whole course.