TL;DR: there are three ways to extract subtitles from a YouTube video, and which one works depends on what kind of subtitles it has. If the video has a caption track (a CC button in the player), copy the text from YouTube's transcript panel or download it as an SRT file with the free tool yt-dlp. If the subtitles are burned into the picture itself, no downloader can extract them — you need OCR software. If the video has no subtitles at all, transcribe the audio with speech recognition. This guide covers all three, starting with the methods that cost nothing and need no software.
Which kind of subtitles does the YouTube video have?
Thirty seconds of checking saves you from using the wrong method. Look at the video player:
| What you see | What it means | How to extract |
|---|---|---|
| CC button — subtitles turn on and off | Caption track (auto-generated or uploaded) | Transcript panel or yt-dlp |
| Subtitles always visible, no CC button | Burned into the picture — no track to download | OCR |
| No CC button, no visible subtitles | The video has no subtitles | Transcribe the audio |
Downloaders and online subtitle extractors only work on the first case — videos with an actual caption track. Burned-in subtitles look identical to viewers but are just pixels, which is why they need a completely different method.
How do you copy a YouTube transcript without installing anything?
For grabbing the text of a single video — for notes, a summary, or feeding to an AI — YouTube's built-in transcript panel is the fastest method and works in any browser.
- Open the video on youtube.com (this does not work in the mobile app for all videos).
- Below the video title, click ...more to expand the description, then click Show transcript.
- The transcript opens with timestamps. Click the three-dot menu inside the panel and choose Toggle timestamps if you want clean text.
- Select the text and copy it.
The limitation: the transcript panel gives you plain text, not a subtitle file — there is no way to export SRT timing from it. If you need a real subtitle file with timestamps, use the next method.
How do you download YouTube subtitles as an SRT file?
yt-dlp, a free open-source command-line tool, can download any public YouTube video's caption track as a subtitle file without downloading the video itself. On a Mac, install it with brew install yt-dlp, then:
# See which caption tracks the video has
yt-dlp --list-subs "VIDEO_URL"
# Download the auto-generated English captions as SRT (no video)
yt-dlp --write-auto-subs --sub-langs en --convert-subs srt --skip-download "VIDEO_URL"
# If the creator uploaded their own captions, prefer those — they are usually more accurate
yt-dlp --write-subs --sub-langs en --convert-subs srt --skip-download "VIDEO_URL"
Replace en with any language code the video offers (--list-subs shows them). Various websites offer the same thing through a paste-a-URL form; they come and go and tend to be ad-heavy, so the command-line route is the reliable one.
Two things to expect from auto-generated captions. First, the text is rough: punctuation is often missing, and names, technical terms, and accented speech get transcribed wrong. Second, the timing is word-by-word rolling captions, so converted SRT files often contain overlapping, partially repeated lines that need cleanup before the file is usable as real subtitles.
What if the subtitles are burned into the video (no CC button)?
Many videos on YouTube — music videos, variety shows, lyric videos, re-uploaded clips, anything edited in CapCut with captions — have subtitles rendered directly into the picture. Burned-in subtitles are part of the video image, so no downloader, browser extension, or online extractor can pull them out; the only way to extract them is OCR (optical character recognition), which reads the text from the frames.
- Download the video with yt-dlp or any video downloader.
- Import it into GeekLink on your Mac and run OCR extraction. It detects the subtitle region, reads each frame's text, and merges duplicates into clean, correctly timed subtitle lines.
- Export as SRT, or translate the extracted subtitles first and export a bilingual file.
OCR runs locally on your Mac — the video is not uploaded anywhere, and a whole batch of videos can be processed in one run. For details, see the full guide: How to Extract Hardcoded (Burned-In) Subtitles Using OCR. If the video also has a channel watermark or on-screen text you want excluded, see Extract Subtitles Without the Watermark.
What if the video has no subtitles at all?
If there is no caption track and nothing burned in, there is nothing to extract — you generate subtitles instead. Download the video, import it into GeekLink, and run Whisper speech recognition locally on your Mac. You get timed subtitles you can edit, translate, and export as SRT, without uploading the audio to a cloud service. See Whisper for Video Transcription: Complete Guide.
How do you translate the subtitles you extracted?
Extraction is often step one of a translation job — a subtitle file in a language you or your audience do not speak. An extracted SRT file can be imported straight into GeekLink and translated with AI into 40+ languages, keeping the original timing — no re-transcription needed. Export the result as a translated SRT, a bilingual SRT, or burn the subtitles into the video for platforms that do not support subtitle tracks. Full workflow: How to Translate YouTube Videos.
FAQ
Can I download subtitles from any YouTube video?
Only from public videos that actually have a caption track. Private, members-only, and age-restricted videos may not be accessible to downloaders, and videos whose subtitles are burned into the picture have no track to download — those need OCR instead.
Why does my downloaded YouTube caption file repeat every line?
Auto-generated captions are stored as word-by-word rolling captions, so when they are converted to SRT, consecutive cues overlap and repeat parts of the previous line. Creator-uploaded caption tracks do not have this problem. If you only have the auto track, the file needs cleanup in a subtitle editor before it reads as normal subtitles.
Are YouTube auto-generated captions accurate enough to use directly?
Usually not for publishing. Auto captions typically lack punctuation and misrecognize names, technical terms, songs, and accented speech. They are fine for personal notes; for subtitles you plan to publish or translate, review and edit them first.
How do I extract subtitles that are part of the video image?
With OCR. Burned-in subtitles are pixels, not text, so downloaders cannot see them. GeekLink runs OCR on the video locally on your Mac, merges the per-frame text into timed subtitle lines, and exports SRT — it can also batch-process many videos at once.
Is it legal to extract subtitles from YouTube videos?
Extracting subtitles for personal use — notes, study, accessibility — is generally acceptable. Republishing subtitles or translations commercially requires permission from the copyright holder. Many creators welcome fan translations that expand their audience; when in doubt, ask.