Get a YouTube transcript as Markdown

Paste a YouTube URL into Milldown and get the video's transcript as clean Markdown — ready for notes, summaries, or your knowledge base. Fetched directly from your Mac, no third-party transcript service involved.

How it works

1. Copy the YouTube video URL.
2. Press ⌘LCtrl+L in Milldown and paste it.Milldown recognizes YouTube links and retrieves the transcript rather than the page markup.
3. Copy or save the transcript as Markdown.Add front matter to keep the source URL attached to your notes.

What people use it for

• Summaries — paste an hour-long talk's transcript into Claude and ask for the five key points.
• Research — quote talks precisely instead of paraphrasing from memory.
• Knowledge bases — conference talks and tutorials become searchable text in your vault.
• RAG pipelines — chunk the transcript (Refine → Export Chunks) and index it alongside your documents.

What you get

The video title, its description, keywords and runtime, and the full transcript — as one Markdown document. Milldown reads the captions YouTube already publishes; no third-party transcript service is involved.

Requirements and limits

The video needs captions — either uploaded by the creator or YouTube's auto-generated ones (most videos have them). Auto-captions carry auto-caption quality: proper nouns and technical terms sometimes come out creative. Private and age-restricted videos aren't accessible.

Where the transcript actually comes from

Milldown fetches the captions YouTube has already published for the video. Nothing is downloaded and no audio is transcribed — there is no speech recognition in this path at all, which is why it is fast and why it costs nothing.

That has one consequence worth knowing before you rely on it: if a video has no captions, there is no transcript to fetch. You still get the title, description, keywords and runtime, but the body will be empty. Most substantial channels have captions, either written or auto-generated, and auto-generated ones work here too.

Language and accuracy

• English is preferred where it exists, otherwise the first published caption track is used. A video captioned only in German returns German.
• Auto-generated captions carry auto-generated errors. Names, technical terms and figures are where they go wrong, and they arrive without the punctuation a human captioner adds. Check any number you intend to quote.
• No speaker labels and no timestamps in the converted text — the transcript arrives as continuous prose. For an interview with several voices, that is a real limitation.

Fair use, briefly

A transcript is the video’s words, and those words belong to whoever made it. Summarising, quoting and taking notes for your own work is ordinary use; republishing a transcript wholesale generally is not. Worth a thought before pasting one somewhere public.

Further reading: Converting a web page to Markdown · Preparing documents for RAG