YouTube Transcript to Markdown
Save the subtitles as a note you can keep

What the reading view does before you clip anything

Open a video and the extension goes looking for the caption track on its own. You do not press anything. It lays the transcript out as readable text instead of a cramped sidebar, split into roughly 30 second blocks, or by the video chapters when the uploader set them. Each block is stamped with a time.

From there it is a document you can work with:

• Click a timestamp to jump the video to that second. The step you are reading and the moment it happens are one click apart.
• The block being spoken right now stays highlighted, and the transcript scrolls itself to follow. A three-state toggle controls that, so you can let it follow the video or stop the scrolling and read at your own pace.
• Ctrl+F is intercepted and searches the transcript instead of opening the browser find bar, so a word said once at 34:12 is findable.
• The language bar lists every caption track the video has, translated tracks included, and switches between them.
• Clicking through to the next video re-extracts the new transcript on its own, since YouTube never reloads the page.

Sometimes YouTube blocks the request with its PO Token anti-scraping check. The extension then falls back to the HTML5 textTracks API on the video element and reads the captions from there.

Which videos are worth saving this way

A transcript is a wall of spoken words. That pays off for some videos. For others it is clutter you will never open again.

The rough test: if you can picture yourself searching for something the speaker said, save it. Otherwise skip it. Watching is enough.

Saving one, step by step

• Install the extension in Chrome, Edge, or Arc. It does not ask to read every site you visit at install time.
• Open a YouTube video that has captions. Press Alt+R, or click the extension icon in the toolbar and choose the reading view.
• Wait for the transcript to load. It arrives already split into blocks. Each one carries a timestamp.
• If the video has several caption languages, pick the one you want in the language bar. That choice is what gets saved.
• Click the clip button in the reading view header. On an ordinary web page that button clips the article text the reading view parsed out. On a video page, it clips the transcript.
• NoteLoom opens. It shows a Markdown preview of what you clipped.
• Pick a local folder. The transcript is written there as a .md file you can open in any Markdown editor, now or in five years.

Every step after the first is a click. Nothing gets highlighted and pasted by hand.

What you actually get, and what you do not

The file is a verbatim record of speech. Someone said "so, um, basically what I do here is" and that is what lands in the .md. NoteLoom does not summarize it or shorten it.

So the file is a starting point, not a finished note. The work left over:

• Cut the filler and the throat-clearing. Most of a transcript is not worth keeping.
• Add your own headings. A timestamp tells you when something was said, and nothing about whether it mattered.
• Write down why you saved it, in your words, before you forget.

Auto-generated captions have their own tax. Names come out wrong, jargon comes out wronger, and punctuation is mostly guesswork. If the speaker mumbles, the transcript records the mumbling faithfully. Human-written captions are much cleaner. Plenty of channels do not bother with them.

And a video with no caption track gives you nothing. The extension reads what YouTube has; it does not listen to audio.

Where the text goes

The caption track is fetched from YouTube's own endpoint, by a content script running inside the page, using the cookies your browser already has for youtube.com. That authentication is why the extension asks for cookies permission at all. The conversion from the reading view content to Markdown runs in your browser, and the file is written by the browser into the folder you granted access to.

No NoteLoom server is involved in that round trip. There is no account to make, and no transcript sitting in someone else's database. The extension source is open for anyone to check.