Burned-In Captions vs Closed Captions Explained

Burned-in captions are text rendered permanently into the video frames, always on, can't be switched off, identical on every device. Closed captions are a separate track the player loads at playback, so a viewer can toggle them on or off and resize them. Same words; one lives in the picture, the other rides alongside it.
That single difference, inside the pixels versus attached to the file, decides almost everything else: whether a search engine can read the words, whether a viewer can turn them off, and whether the captions survive when someone downloads and reposts your clip. Get the terminology straight first, then the when-to-use rule writes itself.
What's the difference between burned-in and closed captions?
Burned-in captions are fused into the image during export. The text becomes part of every frame, like a logo or a watermark, so it shows up no matter what player opens the file and no matter how the clip gets reshared. There's no off switch because there's nothing separate to switch off, the words are pixels now.
Closed captions are stored in their own layer: an SRT or VTT file, or a platform's own caption track. The video player reads that layer at playback and draws the text over the picture on the fly. Because the captions are data rather than image, a viewer can toggle them, a screen reader can announce them, and a search index can read the words. The trade-off is that the layer is detachable, and detachable things get detached.
Open, closed, hardcoded, what do these words actually mean?
They collapse into two buckets. Open captions, burned-in captions, hardcoded subtitles, and hardsubbed all name the same thing: text baked into the frame that's always visible, the last two are the fansub and video-encoding terms. Closed captions, soft captions, and softsubs all name the toggleable kind stored on a separate track. Two ideas, six words.
One footnote worth knowing: in broadcast and accessibility law, "captions" technically include non-speech audio cues (a door slam, music swelling) while "subtitles" assume the viewer can hear and just need the words. On social platforms almost everyone uses the terms interchangeably, so for clip work, treat caption and subtitle as the same job.
When should you use each one? (the rule)
Use burned-in captions for autoplay short-form, Reels, TikTok, Shorts, and other vertical feed video, because those plays start muted and the captions need to be visible the instant the clip appears, no tap required. Use closed captions where viewers actively choose to watch and where accessibility toggles and search indexing matter: a YouTube long-form upload, a course player, an embedded video on your site.
The muted-feed half of that rule isn't a hunch. Publishers told Digiday back in 2016 that roughly 85% of Facebook video was watched with the sound off, publisher-reported data, not a platform audit, and later studies land lower, so treat 85% as the top of a range rather than a fixed law. The direction has held for a decade. On any feed where the clip plays itself, words on screen aren't a nice-to-have; they're whether the silent majority gets the point at all. Burned-in is the only caption that's guaranteed to be there.
The closed-caption half is about two things the pixels can't give you. First, a screen-reader user and anyone who wants larger text needs a track they can control, burned-in text is fixed at one size forever. Second, search engines and platform recommendation systems can read a real caption track but cannot read text painted into an image; to a crawler, burned-in words are just shapes. On YouTube specifically, that's why serious creators do both: burned-in styling for the watch experience, plus an uploaded transcript or caption file so the words are indexable.
This matters more for clips than for full episodes because clips travel. A clip's whole purpose is to get reshared, reposted, and saved, and most of those moves strip a separate caption track but can't touch burned-in text. Short-form clipping has become one of the fastest-growing discovery channels for long-form shows, so every re-upload is another chance for a fragile track to fall off in transit. (For which captions survive on each specific app, see the deeper guide on burned-in vs soft captions for clips.)
Examples: which one shows up where
You've seen both without naming them. The big, bouncing, word-by-word text on a TikTok or a Reel is burned-in, it's there the moment the clip autoplays, sized and styled by the creator, and it stays put when the clip gets reposted. The thin grey caption bar you toggle on a Netflix title or a YouTube video using the CC button is closed, a track the player draws over the picture that you can turn off or restyle.
For podcast work, the split is clean. Short vertical clips for the feed get burned-in captions, full stop. An audiogram, a still or waveform with audio, almost always uses burned-in text, since it's built to be posted and reshared. A quote card bakes its words in by definition; the text is the design. And a full-length YouTube episode gets a real caption file uploaded so the transcript is searchable, often alongside burned-in lower-thirds for emphasis.
Frequently asked questions
Are burned-in captions the same as hardcoded subtitles? Yes. Burned-in, open, hardcoded, and hardsubbed all describe text rendered permanently into the video frames so it's always visible and can't be turned off. Closed, soft, and softsubbed describe the toggleable kind stored in a separate track. The concept is binary; only the words multiply.
Do burned-in captions hurt accessibility? They help the muted-feed majority but limit control: a viewer can't enlarge them or swap them for another language, and a screen reader can't read pixels. For a video where accessibility toggles matter, a site embed, a course, pair burned-in text with a real closed-caption track, or use closed captions alone.
Can search engines read burned-in captions? No. To a crawler or a platform's recommendation system, burned-in text is just part of the image, shapes, not words. Only a real caption track (SRT/VTT) or an uploaded transcript is machine-readable. That's the main reason to add closed captions to a YouTube long-form upload, even when the clip itself is burned-in.
Which should I pick if I only do one? For short-form podcast clips, pick burned-in. The feed autoplays muted, captions have to be visible instantly, and burned-in text is the only version guaranteed to survive a repost. Add a closed-caption track on top only where the words need to be searchable or independently toggleable, mainly YouTube. (Whichever you choose, review the AI-generated captions before posting, auto-captions get names and numbers wrong.)