How to Get the Transcript of a YouTube Video: 4 Ways

A transcript turns a video into text you can skim, search, quote and paste into other tools. Which way of getting one is best depends on what you want to do with it: read along once, keep a copy, process hundreds of videos in code, or let an AI assistant work with the video for you. Here are four options, from the one built into YouTube to the ones built for developers, with what each is good and bad at.

One thing applies to all four. Every method here reads the captions YouTube already has for the video, either uploaded by the creator or generated automatically by YouTube’s speech recognition. If a video has captions turned off, none of them can produce a transcript, and the quality of the text is the quality of those captions.

1. YouTube’s built-in transcript panel

YouTube shows the transcript of most videos right on the watch page, on desktop:

1. Open the video on youtube.com.
2. Click …more to expand the description below the video.
3. Scroll to the end of the description and click Show transcript.

The transcript opens in a panel next to the video, with a timestamp on each line. Clicking a line jumps the video to that moment, and the panel lets you switch between caption languages when the video has more than one. You can hide the timestamps from the panel’s menu.

Good for: reading along or finding a moment in a video you are already watching. It is free, official and needs nothing installed.

Not so good for: keeping the text. There is no download button, so you have to select the whole panel and copy it, which is fiddly on long videos and drags the timestamps along unless you hid them first. It also doesn’t help with anything beyond reading. YouTube moves this button from time to time, so if you can’t find it, look in the description area or the menu under the video.

2. A transcript tool like QuickTube

A web tool fetches the captions for you and gives them back as a clean document. With QuickTube’s transcript generator:

1. Copy the video link. Copy the YouTube URL from the address bar or the Share button.
2. Paste it into QuickTube. Paste the link on quicktube.ai and press Fetch. No account is needed.
3. Read, search or download. Open the Transcript tab to read the timestamped transcript, search it, or download it as a text or subtitle file.

The transcript tab shows every line with its timestamp and has a search box that filters to the lines mentioning a word. You can download the transcript as plain text or Markdown, or as an .srt or .vtt subtitle file that keeps the timing (see downloading YouTube subtitles). Fetching transcripts this way is free and works without an account.

Good for: keeping a copy, quoting, pasting into notes or an AI chat, and anything you want to do next with the video. Because the transcript is already loaded, QuickTube can also summarize the video, pull out its best moments, write a quiz or answer questions about it. Those AI features use credits.

Not so good for: large numbers of videos. Pasting links one at a time doesn’t scale past a handful; QuickTube’s batch processing takes up to 20 at once, and beyond that you want an API.

3. A transcript API

If you are building something, such as a search index over a channel, a research dataset or a summarization pipeline, you need transcripts as data. There are three routes.

The official YouTube Data API has a captions resource, but downloading a caption track requires OAuth authorization and in practice works for videos you own or manage, not for any public video. It is the right choice for your own channel and the wrong one for everything else.

Open-source libraries such as the Python package youtube-transcript-api read the same captions the watch page uses. They are free and work well on a laptop. In production, YouTube often blocks requests from cloud servers, so you end up managing proxies, retries and caching yourself.

A hosted API such as the QuickTube transcript API handles that for you. One POST request with your API key returns the timestamped segments, the full text and the caption language as JSON:

curl -X POST https://quicktube.ai/api/transcript \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $QUICKTUBE_API_KEY" \
  -d '{"video_url": "https://www.youtube.com/watch?v=VIDEO_ID"}'

Each transcript costs 1 credit, and a free account comes with enough credits to try it properly. The developer guide to YouTube transcript APIs compares the three routes in more detail, with Python code.

4. Let an AI agent fetch it (MCP)

If you work with an AI assistant such as Claude Code, Claude Desktop or Cursor, you may not want the transcript at all: you want the assistant to read the video and do something with it. Most assistants can’t open YouTube videos on their own. The Model Context Protocol (MCP) lets you plug in tools that can.

The QuickTube MCP server is a small Python package, quicktube-mcp, that you add to your client with your API key. After that you can ask things like “get the transcript of this talk and write up the three main arguments” and the assistant calls the transcript tool itself. The same server can summarize, find highlights, generate quizzes and fact-check, at the same credit prices as the API.

Good for: people who already work in an AI assistant and want videos to be another source it can read. Not so good for: a one-off read, where the web tool is quicker to reach.

Which one should you use?

• Just reading along? Use YouTube’s transcript panel.
• Want the text to keep, search or download? Use a YouTube to text converter such as QuickTube.
• Processing many videos in code? Use a hosted transcript API, or an open-source library if you can handle YouTube’s blocking yourself.
• Working in an AI assistant? Add an MCP server and ask it about the video directly.

A note on accuracy and reuse

Automatic captions are good on clear speech in widely spoken languages and weaker on names, jargon, accents and crosstalk. If you plan to quote a transcript, check the quote against the video at its timestamp. And a transcript is still the creator’s content: quoting and studying it is one thing, republishing it wholesale is another, so stay within YouTube’s terms and the creator’s rights.