Captioning for deaf and hard-of-hearing viewers
By Captionwale · 31 October 2026
Deaf and hard-of-hearing viewers need more than words: who speaks, key sounds, exact wording, timing and contrast. What to add, and how to do it in Captionwale.
Captions as the soundtrack
Lots of short-video captions are written for people scrolling on mute. For deaf and hard-of-hearing viewers, though, the captions stand in for the whole soundtrack. A bare transcript drops things a hearing viewer notices without trying: the identity of the speaker, a laugh that tells you it was a joke, music that makes a scene tense, a knock that makes everyone on screen turn.
Captions that work for these viewers carry every sound that changes the meaning, and carry it accurately. No special tool is needed. What it takes is one careful pass through the video.
Cues to add beside the speech
Add these wherever the video needs them. Keep each cue in the caption's own language and script; a Devanagari transcript shouldn't drop into English each time a door closes.
- Who is talking, when the shot doesn't make it obvious: an off-camera voice, a narrator, two people in the same shot. Lead with the name, for example 'Riya: chalo, shuru karte hain', and start a new caption each time the speaker changes.
- Meaningful sounds and music, in square brackets, such as [laughter], [door slams], [music stops] or [soft piano]. Include lyrics if the video leans on them. Leave out sounds that add nothing, or the cues will crowd out what people say.
- How a line is delivered, if the words alone would mislead: open it with [whispering] or [sarcastic].
- What was really said. Smoothing over a stammer is up to you. Summarising, cutting a joke's setup or softening someone's wording changes the message the viewer gets.
Pace, movement and contrast
Captions are read in full by anyone depending on them, so give every one enough screen time to be finished. The 'Min duration' control in Caption settings keeps brief captions up longer. Styles that reveal one word at a time are hard work, because the sentence never shows in full; when meaning matters, a phrase template or a stacked one is kinder.
Captions should land with the speech. For how to check that, see captions out of sync. Words are also harder to read while they move, so it is worth choosing None for Entrance in the Motion pane.
On bright or busy footage, pale letters need a dark rim or a solid panel behind them. If the template allows it, turn on 'Drop shadow', 'Text stroke' or 'Background' in the Effects pane. Some viewers lip-read as they read, so don't cover the speaker's mouth with captions. Don't let colour be the only signal for anything either, since not everyone can tell colours apart.
Open captions versus closed captions
Open captions, the burned-in kind, are part of the picture itself. Everyone sees them, on every app and device, looking exactly as you designed them. The cost: nobody can hide or enlarge them, screen readers get nothing from them, and a fix means exporting and uploading all over again.
Closed captions live in a separate text track that appears only when a viewer asks for it. A caption file supplies them. On paid plans, Captionwale downloads two kinds: SRT, which YouTube takes and which Premiere, Resolve and similar editors import, and VTT, which web-page video players use. Viewers can toggle them, and some players offer a choice of size and colour. Your fonts, colours, placement and animation are not stored in either file.
Offer both and anyone who enables closed captions on a burned-in video reads everything twice. The files also mirror the grouping on screen. Downloaded while a One word template is active, the file has one word per cue, so move to a Single line template before downloading.
Your part and Captionwale's part
Captionwale converts speech to text. Naming speakers and describing sounds is left to you, and whatever you add also goes into both caption files, SRT and VTT. There are two methods:
- A sound between lines: '+ Line' (top of the Captions panel) inserts a caption you write yourself. It lands at the playhead, or else in the next silence wide enough for it. It is labelled TYPED, its In and Out times are yours to set, and re-syncing to the audio leaves it alone.
- A name, or a sound heard during speech: double-click the caption's opening word, or use 'Edit text' on it, then type the cue ahead of it. Because the word keeps its timing, the cue appears at the same moment.
Limits of this method
Captionwale can't judge that a slammed door drives the plot, or recognise an off-screen voice as a particular person's. And when it mishears one real word as another real word, nothing in the caption list looks wrong. Check every caption against the audio, start to end.
Putting a cue inside a word is a work-around; there is no dedicated cue feature. Since only one caption is on screen at once, a typed caption cannot sit over speech. A cue inside a word also behaves like that word: a One word template shows it only while the word is up, and spoken-word highlight styles light the cue up along with it.
This is common accessibility practice. It is not a legal requirement or a broadcast standard, and some clients and platforms require more. The most useful test is feedback from someone who depends on captions, pointing out where they lost track. Without that, watch with the sound muted and mark each spot where you were guessing.
Questions
Do automatic captions do the job for deaf viewers?
Not alone. They record speech, but not who is speaking or which sounds matter, and they can mishear.
How do I write a sound effect into captions?
Describe it briefly inside square brackets, like [door slams], at the moment it happens. In Captionwale, a sound in a pause goes in with '+ Line'; a sound during speech gets typed before a word.
Is a burned-in caption the same thing as a closed caption?
No. A burned-in caption is part of the image and always visible. Closed captions sit in their own file, SRT or VTT for example, and viewers switch them on or off.
Should the captions stay in the spoken language or be translated to English?
If your viewers can read the language being spoken, captions in its own script or in Latin letters preserve exactly what the speaker said. For viewers who don't know that language, Captionwale's English translation carries the meaning instead.