Key takeaways
- Captions show speech and meaningful sounds on screen for people who cannot hear the audio; subtitles translate dialogue for people who do not know the language.
- WCAG 2.2 requires captions for prerecorded video with audio at Level A and for live video at Level AA, and a text alternative such as a transcript for audio-only content at Level A.
- UK public sector websites must meet WCAG 2.2 AA, although pre-recorded audio and video published before 23 September 2020 and live audio and video are listed as things that may not need fixing.
- GOV.UK publishing guidance calls a transcript good practice for video and says audio-only recordings such as podcasts need one.
- Translated subtitles are most accurate when produced from a checked source-language transcript rather than from automatic captions.
Marketing teams, public sector communicators, e-learning producers and event organisers all face the same question when they publish video or audio: do we need captions, subtitles, a transcript, or all three? The terms are often used interchangeably, but they do different jobs for different audiences, and accessibility rules treat them differently.
This article defines each format, compares them in one table, sets out what WCAG 2.2 and UK public sector rules require, and explains how translated subtitles are produced from a transcript. For the wider subject, see our guide to transcription and translation of audio and video.
What is the difference between subtitles, captions and transcripts?
Captions, subtitles and transcripts are all text versions of spoken content, but captions and subtitles are timed to appear on screen with the video, while a transcript is a separate document. Captions serve people who cannot hear the audio; subtitles serve people who cannot understand the language.
- Captions are on-screen text of the speech and the meaningful sounds in a video. The W3C's guide to captions defines them as "a text version of the speech and non-speech audio information needed to understand the content."
- Subtitles are on-screen translations of the dialogue for "people who can hear the audio but do not know the spoken language", in the W3C's words. The same page notes that "Captions are needed for accessibility, whereas subtitles in other languages are not directly an accessibility accommodation."
- Transcripts are text documents of the content. The W3C's guide to transcripts describes basic transcripts as "a text version of the speech and non-speech audio information needed to understand the content", while "Descriptive transcripts for videos also include visual information needed to understand the content."
Usage varies by country. The W3C's guidance on Success Criterion 1.2.2 notes that "In some countries, captions are called subtitles", and in UK broadcasting "subtitles" commonly covers both, with the deaf-oriented version often called subtitles for the deaf and hard of hearing (SDH).
| Feature | Captions | Subtitles (translated) | Transcript |
|---|---|---|---|
| Main audience | Deaf and hard of hearing viewers; anyone watching without sound | Viewers who do not understand the spoken language | Deaf, deafblind and other users; readers who prefer text; search engines |
| Language | Same as the audio | Different from the audio | Usually same as the audio; can be translated |
| Timed to video | Yes | Yes | No (may include timecodes) |
| Non-speech sounds | Included, with speaker identification | Usually dialogue only | Included |
| Visual information | No | No (except translated on-screen text) | Included in a descriptive transcript |
| Typical format | SRT or WebVTT file, or burned in | SRT or WebVTT file, or burned in | HTML page or document |
| WCAG 2.2 role | Required for prerecorded (Level A) and live (Level AA) video with audio | Not an accessibility requirement | Satisfies Level A for audio-only content |
What do accessibility rules require for video and audio content?
Accessibility rules based on WCAG 2.2 require captions for video with sound and a text alternative, such as a transcript, for audio-only content. Translated subtitles are not required for accessibility.
The key WCAG 2.2 success criteria are:
- 1.2.1 Audio-only and Video-only (Prerecorded), Level A. For audio-only content, the criterion requires "An alternative for time-based media" that "presents equivalent information", which in practice means a transcript.
- 1.2.2 Captions (Prerecorded), Level A. "Captions are provided for all prerecorded audio content in synchronized media". The Understanding document adds that captions "identify who is speaking" and "include non-speech information conveyed through sound, including meaningful sound effects."
- 1.2.4 Captions (Live), Level AA. The criterion reads: "Captions are provided for all live audio content in synchronized media."
UK public sector websites and apps
Public sector bodies must meet the accessibility regulations. GOV.UK's accessibility requirements for public sector websites and apps say sites must "meet the Web Content Accessibility Guidelines (WCAG) 2.2 AA accessibility standard". The same guidance lists, among things you might not need to fix, "pre-recorded audio and video published before 23 September 2020" and "live audio and video". New pre-recorded video published by a public body therefore needs captions, and audio-only content needs a transcript.
Private sector organisations
The public sector regulations do not apply to most private companies, but WCAG 2.2 AA is the standard most organisations use to judge accessibility. Captioning marketing and training video also reaches the many people who watch with the sound off.
Do you need a transcript as well as captions?
A transcript is not strictly needed for video that already has accurate captions, but it is good practice and it serves people captions cannot. For audio-only content such as podcasts, a transcript is the accessible alternative.
The GOV.UK content and publishing guidance on videos says:
"Although a transcript is not necessary for meeting accessibility rules, it’s good practice to provide one."
The same guidance says "Audio-only recordings, like podcasts, also need a transcript", and that a transcript should "include all important audio and visual information in the video". Its example is a press conference ending with a handshake between ministers that signals a deal: if no one mentions it aloud, the transcript has to describe it. Transcripts also "help search engines and AI assistants find the content", which is a practical benefit for marketing teams.
Are automatic captions good enough?
Automatic captions are a useful first draft, but they are rarely accurate enough to publish without review. The W3C's captions guidance says of automatic captions: "Usually they need significant editing."
GOV.UK's publishing guidance asks editors to go through automated captioning and check that it "is in sync with the person speaking", that "there are no spelling mistakes", and that it "has captured the correct words, paying special attention to uncommon words or phrases". Names, product terms, acronyms and accents are where automatic captions fail most often, and automatic captions in one language are a weak basis for translated subtitles.
How much subtitling does UK broadcast television require?
UK broadcast television has statutory subtitling targets. Under section 303 of the Communications Act 2003, the obligations from the tenth anniversary of the relevant date require at least 90% of Channel 3 and Channel 4 programming and at least 80% of other services within scope to be subtitled, with at least 10% audio-described and 5% presented in or translated into sign language.
These targets apply to broadcasters and are enforced through Ofcom's code on television access services. They do not apply to company websites or social media, but they show the level of provision that UK audiences have come to expect.
How are translated subtitles produced from a transcript?
Translated subtitles are produced by transcribing the original audio, timing it into caption segments, then translating and adapting each segment to fit on screen. Working from an accurate source transcript is what keeps the translation faithful and the timing right.
- Transcribe the audio in the original language, usually in clean verbatim.
- Segment and time the transcript into caption units and export a timed file such as SRT or WebVTT.
- Translate each segment, condensing where needed so viewers can read it in the time available, since reading is slower than listening.
- Translate on-screen text such as titles, captions in graphics and signs that matter to the story.
- Review in the player to check timing, line breaks, shot changes and readability on mobile and desktop.
Where the source audio is itself in another language, start with a source-language transcript; see foreign-language transcription and translation. Game studios should also read about voice-over and subtitling for games.
Which format should you choose?
The right format depends on who must be able to access the content and in which languages. Use this checklist:
- Video with speech, published by a public body: accurate captions, plus a transcript as good practice.
- Audio-only content such as podcasts: a transcript.
- Video for international audiences: same-language captions plus translated subtitles for each target market.
- Training and e-learning video: captions and a downloadable transcript, translated where learners need it (see e-learning localisation).
- Meetings, interviews and research recordings that are not published: a transcript only; see transcribing research interviews.
Common mistakes with subtitles, captions and transcripts
Most problems arise when one format is expected to do another's job. Avoid these mistakes:
- Publishing translated subtitles as "captions". Dialogue-only subtitles leave out speaker identification and sound information that deaf viewers need.
- Burning in subtitles when a file would do. Open (burned-in) text cannot be switched off, resized or replaced with another language.
- Uploading unedited automatic captions. Errors in names and key terms undermine credibility and accessibility.
- Leaving visual information out of transcripts. Diagrams or actions that are not described aloud are lost to people who cannot see the video.
- Translating from automatic captions. Errors in the source multiply in every target language.
How TranslationWorks can help with transcripts and subtitles
TranslationWorks is a UK-based translation agency that transcribes audio and video and translates the transcripts into 150+ languages, so the text can be used for transcripts, captions or translated subtitles. The work is done by native-speaking linguists, and TranslationWorks is ISO 17100 certified, ISO 9001 and ISO 27001 registered and Cyber Essentials certified, which matters for unreleased or internal video. Tell us the video length, languages and the file format your player or platform needs. See our transcription services and translation services, or get a free quote, which we send within 24 hours.


