Captions and subtitles are not quite the same thing
The two words are often used interchangeably, but they started as different ideas. Subtitles generally assume the viewer can hear the audio and need the words translated or transcribed. Captions assume the viewer may not be able to hear, so they also include meaningful non-speech sound, such as music cues, laughter or a door slamming off screen, and they often identify who is speaking when that is not obvious.
In practice, many platforms use “subtitles” and “captions” loosely. What matters is deciding what your text needs to do. A viewer who is deaf or hard of hearing needs the fuller, caption-style information. A viewer who is watching in a second language mostly needs an accurate translation of the speech. Many people also simply watch with the sound off in public or on a phone, and benefit from either.
Open and closed captions
There are two basic ways to deliver text with a video.
- Open captions are burned into the picture. They are always visible, they cannot be turned off, and they look the same everywhere the video is played. The downside is that you cannot change them later without re-exporting, and they cannot be switched to another language.
- Closed captions travel separately from the picture, as a text track the viewer can switch on or off. The player draws the text, so the viewer may be able to adjust its size or color, and you can offer several languages from one video.
Closed captions are usually the more flexible choice. Open captions can make sense where a platform does not support text tracks well, or where you want precise control over the look, as is sometimes the case with short clips for social feeds.
Step 1: Start with a transcript
Good captions begin with an accurate script of what is said. If you wrote a script, you may be able to start from it, but check it against the recording, because people rarely say exactly what was written. Otherwise, you can transcribe by listening, or use a speech-to-text tool as a first draft.
Automated transcripts are a useful time saver, and tools for automated editing increasingly include them, but they are not reliable enough to publish unchecked. Names, specialist terms, accents and overlapping voices are common trouble spots. Plan to read through the whole transcript while listening.
Step 2: Break the text into caption cues
A caption cue is one block of text that appears on screen for a set time. Breaking text into cues is mostly a readability exercise.
- Keep each cue short. Two lines at most is a widely used convention, with each line kept short enough to read quickly. Many guidelines suggest roughly 32 to 42 characters per line, though recommendations differ between platforms and broadcasters.
- Break lines at natural phrase boundaries rather than in the middle of a phrase. “She went to / the store” reads worse than “She went to the store / after work”.
- Avoid splitting a name, a number or a verb phrase across two cues where possible.
- Keep punctuation conventional so sentences are easy to parse.
Step 3: Timing
Each cue needs a start time and an end time. The simplest rule is that text should appear when the words begin and disappear shortly after they end. A few practical points:
- Leave a cue on screen long enough to be read comfortably. Very short cues are hard to read even when the words are few. Reading speed varies a lot between people, so err toward more time rather than less.
- Where possible, do not let a cue linger across a hard cut or scene change if the speaker changes, because it can feel like the words belong to the wrong shot.
- Keep a small gap between consecutive cues so the eye registers that the text has changed.
Most caption tools show the waveform or a video preview so you can nudge the start and end of each cue by fractions of a second. Spend extra time on fast dialogue, where timing drifts most easily.
Step 4: Choose a caption file format
If you use closed captions, you will export or upload a caption file. Two common plain-text formats are:
- SRT (SubRip). Each cue has a number, a time range written like
00:00:01,000 --> 00:00:03,500, and one or two lines of text. It is simple and widely supported, but it carries very little styling information. - WebVTT. It starts with the line
WEBVTT, uses a similar time-range line (with a full stop rather than a comma before the milliseconds) and can include positioning and some styling. It is commonly used with video on web pages.
Other formats exist, especially in broadcast. Which one you need depends on the platform or player, so check what it accepts. Because these files are plain text, you can open one in a text editor to fix a typo, which is often quicker than going back into the editing software.
Step 5: Style for readability
Style matters most with open captions, but it can also apply to closed captions where the player allows it.
- Use a clear, plain typeface in a reasonably large size.
- Make sure there is strong contrast between the text and the background. White text with a dark outline or a semi-transparent box behind it is a common solution, because the footage behind the text changes constantly.
- Keep captions inside the area where the screen or app interface will not cover them. On a phone, buttons and descriptions often sit near the bottom edge, so some creators lift the text slightly. This matters particularly for mobile viewing, where screen sizes vary.
- Be consistent across the whole video, including placement, color and how you label speakers.
Step 6: Check before publishing
Play the video from start to finish with the captions on, ideally once with the sound off to experience what a silent viewer would see. Check for typos, mistimed cues, text running off-screen and sound descriptions that are missing. If you are providing translations, have a fluent speaker review them if you can, since a literal translation may miss tone and idiom.
Why this is worth the effort
Captions help people who cannot hear the audio, people in noisy or quiet places, and people who find reading easier than listening. They are also a common part of video accessibility practice. Whether captions are legally required depends on the country, the type of organization and the content, so check the rules that apply to you rather than assuming.
They also help viewers who use video for language learning, since reading and hearing the same words together can support comprehension, although how much it helps varies from person to person. Captions are a different thing from audio description, which describes the visuals for people who cannot see them; the two address different needs and can be offered together.