Closed Captions vs Subtitles: What Is the Difference?

Closed captions and subtitles both put text on screen, but they solve different problems. Captions are written for viewers who cannot hear the audio. Subtitles usually assume the viewer hears the sound but does not understand the spoken language. This guide explains the difference for recorded videos, streams and video calls, and helps you decide which one your content needs.

What closed captions and subtitles are

Closed captions transcribe the dialogue and also carry other audio information: sound effects, music cues, speaker identification and sometimes tone, such as [door slams] or [whispering]. "Closed" means the viewer can switch them on or off. The opposite is open captions, which are burned into the picture and always visible.

Subtitles normally show only the spoken dialogue and relevant on-screen text, often translated into another language. They assume the viewer can hear music, effects and who is speaking. Like captions, subtitles can be a selectable track or burned into the video.

Terminology varies by region and platform. In the UK and many other countries, "subtitles" is used for both types. There is also a hybrid called SDH (subtitles for the deaf and hard of hearing), which adds non-speech information to a subtitle track. When a platform, client or regulator asks for one or the other, check their specification rather than relying on the label.

Key differences between captions and subtitles

• Audience: captions serve deaf and hard-of-hearing viewers and anyone watching without sound. Subtitles serve viewers who hear the audio but need a translation or prefer to read along.
• Content: captions include non-speech sounds and speaker labels. Subtitles generally stick to dialogue.
• Language: captions are normally in the same language as the audio. Subtitles are often a translation.
• Visibility: closed captions and most subtitle tracks can be toggled by the viewer. Open captions and burned-in subtitles cannot be turned off.
• Placement: caption guidelines often call for moving text so it does not cover important visuals and for marking off-screen speakers. Exact requirements differ by broadcaster, platform and regulation.

Selectable text usually travels as a separate sidecar file or as a track embedded in the video container. Common formats include SRT and WebVTT, with SCC and TTML used in broadcast and some streaming workflows. Which formats are accepted, and whether styling and positioning are kept, depends on the platform, player and version, so check the current upload documentation. Burning text into the image makes it appear wherever the file plays. The trade-off is that the text cannot be edited, resized or switched off without re-exporting, and players and assistive tools cannot read it as text.

Many meeting and streaming apps offer automatically generated live captions, and some also offer translated captions. These rely on speech recognition, so accuracy depends on audio quality, microphone placement, background noise, overlapping speakers, accents and specialist vocabulary. Products, plans and settings differ on three privacy points: whether audio is processed on the device or by a remote service, whether transcripts are saved, and who can access them. Confirm these in the provider's current documentation and your organization's policy before relying on live captions in confidential meetings.

Practical trade-offs

Accessibility obligations are the clearest dividing line. The Web Content Accessibility Guidelines (WCAG) call for captions on prerecorded audio in synchronized media. Dialogue-only subtitles usually do not meet that expectation because they leave out sounds that carry meaning. If your video falls under education, public-sector or broadcast rules, check the specific standard and jurisdiction that applies.

For social video that autoplays muted, open captions guarantee the text is seen even where the player does not show selectable tracks. However, the style is fixed and viewers cannot hide it. Selectable captions let the platform or viewer adjust size and contrast, but only appear when the player supports them and the viewer turns them on.

Automatic captioning is quick to produce but needs human review. Names, product terms, numbers and quiet or overlapping speech are common sources of errors. Accuracy is something to check on your own recordings, not something to assume from a feature description. A clean, close microphone signal helps any recognition engine, so test your audio before recording or going live.

For multilingual content, a reviewed same-language transcript is a practical starting point for translated subtitles. Translated lines are often longer or shorter than the original, so recheck timing and line breaks for each language instead of reusing the source timing unchanged.

Which one to choose

• Use closed captions if viewers may be deaf or hard of hearing, if accessibility rules apply, or if sound effects and speaker changes matter to understanding.
• Use subtitles if the audience can hear but speaks a different language from the audio.
• Use SDH if you need translation and accessibility in one track, or if your platform labels its accessible tracks as subtitles.
• Use open, burned-in captions if you cannot control the player, as with some social feeds, and you accept that viewers cannot disable or restyle them.
• Use live captions in calls as an aid to following the conversation, and keep a reviewed transcript or caption file when an accurate record is required.

Before publishing, play the final file in the target player with the sound off. Confirm that the correct track appears, that the timing matches the speech, and that the text does not hide important parts of the picture.