How to Get the Script of a YouTube Video (3 Free Ways)

A video breaks out in your niche — a few hundred thousand views in a week — and the useful question isn’t “what did they say,” it’s “why did that exact opening line hold people when mine didn’t.” You can’t answer that by rewatching at 1x with a notepad. You need the words as text, side by side with your own draft. Learning how to get the script of a YouTube video is a five-minute problem, and it’s free — no software, no browser extension you have to trust, no signup — through any of the three routes below.

All three pull from the same place: the caption track YouTube already generated or the creator already uploaded. There isn’t a hidden, more complete transcript one method can reach that another can’t — the difference between them is speed, convenience, and whether you need one video or fifty.

Method 1: YouTube’s built-in transcript panel

This is the fastest option for a single video and it’s built into YouTube itself — no third-party site, nothing to install. YouTube’s own help page confirms it works for “the full transcript of any video that has captions,” which is the one real limitation: no captions, no transcript, from any method.

Method 2: Free transcript extractor tools

Skip the panel entirely by pasting the video’s URL into a transcript-extractor site. These are useful when you don’t want to open the video itself — screening several competitor videos back to back is faster as a list of pasted links than a list of open tabs — or when you want clean, continuous text without scrubbing timestamps out yourself.

1. Copy the YouTube video’s URL.
2. Paste it into a free extractor such as Tactiq’s YouTube transcript tool, which requires no signup and no email to get the text.
3. The tool fetches YouTube’s own caption data and renders it as a page of text.
4. Copy it directly, or use the tool’s download option to save a .txt file.

Method 3: yt-dlp, for more than one video at a time

The first two methods top out at one video, manually, every time. yt-dlp — an open-source command-line downloader with over 184,000 GitHub stars — pulls subtitles directly, without opening a browser, and it’s the only method here that scales to a batch of URLs in a text file or a loop in a script.

1. Install it: brew install yt-dlp on macOS, or pip install yt-dlp anywhere with Python.
2. Run one command, with no video download at all:

   yt-dlp --skip-download --write-subs --write-auto-subs \
     --sub-langs en --convert-subs srt "https://www.youtube.com/watch?v=VIDEO_ID"

3. --skip-download keeps the video itself off your disk. --write-subs grabs creator-uploaded captions if they exist; --write-auto-subs falls back to YouTube’s auto-generated track when they don’t. --convert-subs srt turns the raw caption file into a format any text editor can open cleanly.
4. For a batch, put one URL per line in a file and add -a urls.txt in place of a single URL — yt-dlp works through the whole list.

Is this legal? A fast copyright note

Pulling a transcript to read, study, or take notes from is squarely inside fair use — YouTube’s own guidance names research and commentary or criticism as classic examples, and courts weigh whether a use is transformative — whether it adds new expression or meaning rather than just copying — as the central question.

What creators actually do with a competitor’s script

The transcript by itself is just text. The reason creators bother pulling one at all is to measure three specific things a silent read-through won’t surface on its own: the exact wording of the hook, where the video re-opens attention partway through, and how fast the narrator is actually talking. On X, this has become a genuinely common workflow among faceless-channel operators — pull the transcript of a proven video, feed it to an AI assistant with a prompt built to extract hook pattern, structure, and pacing, then write an original script in the same shape for a different topic, without copying the wording. The consistent framing among people doing this: it’s pattern research, the same as a runner studying another runner’s stride, not plagiarism of the words themselves.

Once you’ve pulled those three measurements from a video that’s working, the actual bottleneck moves to writing your own version at the same discipline — matching the hook style and the pacing without copying a sentence of it. That craft is covered in how to write a video script, including the same loop-and-payoff structure the teardown above is measuring for. Longform Studio’s script step is built for exactly that handoff: you draft against a measured words-per-minute target instead of a guess, and ElevenLabs narration timing updates on every sentence you edit, so a pacing decision made from a competitor’s transcript actually holds once it’s recorded.

This kind of research matters more the longer your videos run — a wpm gap that’s invisible over 90 seconds compounds over 15 minutes, which is part of why runtime and pacing decisions get harder as videos get longer. And if you’re still deciding whether AI-assisted scriptwriting is worth building into your process at all, the criteria for evaluating a YouTube script generator are a separate, specific question from the manual teardown above. Competitor research like this is one piece of the larger system covered in starting a faceless YouTube channel — and once a script is finished, turning it into a rendered video is its own step, laid out in how script-to-video AI actually handles that.