"Turn on the subtitles." "Does this video have captions?" "Can you send me the transcript?" Most people use these words as if they meant the same thing. They don't, and the difference decides whether your video is actually accessible, whether it works in another country and whether search engines and AI tools can read it.
Quick answer
Captions are for people who can't hear the audio. They include the dialogue plus who is speaking and the sounds that matter. Subtitles are for people who can hear but don't understand the language. They translate the dialogue and usually nothing else. Transcripts are the full text of the audio as a document you read on its own, with no timing. SDH (subtitles for the deaf and hard of hearing) is a subtitle track written to caption standards.
| Captions | Subtitles | SDH | Transcript | |
|---|---|---|---|---|
| Main audience | Deaf and hard of hearing viewers | Viewers who don't speak the language | Deaf and hard of hearing viewers | Anyone who wants to read, search or quote |
| Language | Same as the audio | Usually a translation | Same as the audio or translated | Same as the audio (can be translated) |
| Includes speaker labels | Yes, when it isn't obvious | Rarely | Yes | Usually |
| Includes sound cues | Yes, e.g. [phone rings] | No | Yes | Sometimes |
| Timed to the video | Yes | Yes | Yes | No (some include timestamps) |
| Typical file formats | SRT, WebVTT, SCC, embedded 608/708 | SRT, WebVTT, TTML | SRT, WebVTT, TTML | TXT, DOCX, PDF, HTML |
| Required for accessibility | Yes, for video with audio | No | Counts as captions when done well | Yes for audio-only content |
| Main benefit beyond access | Sound-off viewing | International reach | Works where caption formats don't | Search, skimming, reuse |
What captions are
Captions are a text version of everything you would miss if you couldn't hear. That means three things:
- Dialogue, as spoken, including the words of anyone on screen or off it.
- Speaker identification, when it isn't obvious who is talking: a name, a label like NARRATOR, or a position cue for off-screen voices.
- Meaningful non-speech audio: [laughter], [tense music], [glass shatters], [speaking Spanish]. The test is whether the sound changes the meaning or mood. A dog barking in a park scene probably doesn't need a cue. A dog barking at 3 a.m. in a horror film does.
Captions come in two delivery forms. Closed captions can be turned on and off by the viewer and live in a separate track or file. Open captions are burned into the picture and are always visible. We compare those in open captions vs closed captions.
Why captions exist
Captioning grew out of television accessibility. In the US, closed captions on broadcast TV date back to the early 1980s, and the Television Decoder Circuitry Act of 1990 required most new TV sets to include a caption decoder. That's the world the "CC" symbol comes from. Online video inherited the expectation, and the Web Content Accessibility Guidelines (WCAG) make captions for prerecorded video a Level A requirement. Our video accessibility guide covers the standards in detail.
What subtitles are
Subtitles are a translation of the dialogue for viewers who can hear the audio but don't speak the language. The viewer hears the door slam and the music swell, so subtitles don't describe them. They also don't usually label speakers, because the viewer can hear which voice is talking.
Good subtitles are not a word-for-word translation. Reading is slower than listening, so subtitlers condense: they keep the meaning, drop filler and repetition, and fit each line to a comfortable reading speed. That is a craft of its own, and it's why two subtitle tracks of the same film can read quite differently.
Subtitles also include forced narratives (sometimes called forced subtitles): short translations that appear even when subtitles are off, for things like a sign, a text message on screen or a line in a third language that the film's own audience isn't meant to understand either.
What SDH is
SDH sits in between. It's a subtitle track (same file formats, same delivery) written to caption standards: speaker labels, sound cues and music descriptions included.
SDH exists for practical reasons. Some platforms, discs and regions don't support a separate closed caption format, so distributors deliver an SDH subtitle track instead. SDH can also be translated: a French SDH track for a German film serves deaf French viewers, which neither plain captions nor plain subtitles do.
If you're producing one accessible text track for the web, an SDH-style track in WebVTT or SRT is often the most flexible choice.
What a transcript is
A transcript is the complete text of the audio as a document: no timing, no on-screen display, just words you can read, search, copy and quote. A good transcript includes speaker names and, for accessibility, descriptions of important sounds and visual information.
Transcripts serve people captions don't:
- Deafblind users who read text through a braille display. Captions flash by; a transcript can be read at any pace.
- Screen reader users and people who prefer reading to watching.
- Researchers, students and journalists who need to search, quote and cite what was said. If you work with interviews, see how to analyze interview transcripts.
- Search engines and AI tools, which can index and reason over text far more easily than audio.
Some transcripts include periodic timestamps, which is a useful middle ground: still a readable document, but you can jump back to the moment in the video.
US vs UK usage
Part of the confusion is regional:
- In the US, Canada and Australia, "captions" means same-language text for deaf and hard of hearing viewers and "subtitles" means translation. The distinction is common in everyday speech and in regulation.
- In the UK and much of Europe, "subtitles" is the everyday word for both. British broadcasters call deaf and hard of hearing captions "subtitles", and the TV menu option usually just says Subtitles.
Neither usage is wrong. If you're writing a brief for a vendor or a policy document, spell out what you need ("same-language captions with speaker IDs and sound cues") rather than relying on the word.
Which ones do you need?
| Your situation | What to produce |
|---|---|
| Any prerecorded video with speech, for the public | Closed captions (or SDH) in the spoken language |
| Video for an audience that speaks other languages | Captions plus translated subtitles per target language |
| A podcast or other audio-only content | A transcript (captions don't apply without video) |
| Social video that autoplays muted | Captions, often burned in as open captions |
| Lectures, webinars and training | Captions plus a downloadable transcript |
| Research interviews or user tests | A transcript first; captions only if you'll share clips |
| Content you want found in search or used by AI | A transcript published on the page |
A practical rule: start from an accurate transcript. Captions are the transcript broken into timed segments, subtitles are the transcript translated and condensed, and SDH is captions in a subtitle file. If the transcript is right, everything downstream is easier.
Common myths
"Auto-generated captions are good enough." Automatic captions are a useful draft, but they routinely miss names, technical terms and speaker changes, and they never add sound cues. For accessibility, a human should review them.
"Subtitles count as captions." Translated subtitles without sound cues don't meet the caption requirement for viewers who can't hear. SDH can.
"Only deaf viewers use captions." Plenty of hearing viewers turn them on to follow accents, fast speech or noisy scenes, to watch with the sound off, or to learn a language.
"A transcript replaces captions." For video, no. A transcript helps, but a deaf viewer still needs to know what's being said while it's happening on screen.
"Captions and subtitles are just formats." Formats like SRT and WebVTT carry either one. The difference is in what the text says, not the file. See SRT vs VTT for the format side.
How captions, subtitles and transcripts connect
All three start with the same raw material: an accurate record of what was said, with timing. That's the part we work on at YouTubeTranscript.dev. For any YouTube video, it pulls the existing captions or transcribes the audio with AI when there aren't any, then lets you export a plain transcript (TXT), timed captions (SRT or VTT) or a translation into 100+ languages.
If you want to try it on one video, the free transcript tool gives you the text and the subtitle downloader gives you a timed file. For hundreds of videos, the API returns the same data as JSON. Whatever you use, review the result before you publish it as captions: speaker labels and sound cues still need a human eye.
Frequently asked questions
What is the difference between captions and subtitles?+
Captions assume the viewer cannot hear the audio, so they include dialogue plus speaker labels and meaningful sounds like [door slams] or [music]. Subtitles assume the viewer can hear but does not understand the language, so they translate dialogue only. In the UK, 'subtitles' is often used for both.
What does SDH mean on subtitles?+
SDH stands for Subtitles for the Deaf and Hard of Hearing. It is a subtitle track written to caption standards: dialogue plus speaker identification and sound cues. Streaming services use it when a platform or region does not support a separate closed caption format.
Are subtitles and captions the same thing in the UK?+
Mostly, yes, in everyday speech. British broadcasters commonly call deaf and hard of hearing captions 'subtitles', and the term 'captions' is more common in the US, Canada and Australia. The technical distinction between the two still exists in both places.
Is a transcript the same as closed captions?+
No. A transcript is the full text of the audio as a standalone document, without timing. Captions are the same words broken into short timed segments that appear on screen in sync with the video. You can produce one from the other, but they serve different needs.
Do I need both captions and a transcript for accessibility?+
For prerecorded video with audio, WCAG asks for captions. Transcripts are required for audio-only content like podcasts and are strongly recommended for video, because they help deafblind users, people using screen readers and anyone who prefers to read or search.
Why do people watch videos with subtitles on?+
Many hearing viewers turn on captions or subtitles to follow fast dialogue, accents or mumbled lines, to watch with the sound off in public, or to learn a language. That is a big reason captions now matter for reach, not only for accessibility.