Word-by-word captions: why they work and how to do them right
Word-by-word captions show each word as it’s spoken. They work because many people watch with the sound off, because reading along helps follow a fast explanation, and because they make video accessible. Done right, they’re timed to the voice, short, high-contrast and inside the safe zone.
Word-by-word captions (also called animated or dynamic captions) display the narration on screen one word or one short phrase at a time, exactly as it’s spoken, often highlighting the current word. They work for short videos for three simple reasons: a lot of people scroll with the sound off, reading along makes a fast explanation easier to follow, and captions make your video accessible to viewers who are deaf or hard of hearing. To do them right, time them to the actual voice, show only a few words at once, use strong contrast and keep them inside the safe zone.
What word-by-word captions are (and aren’t)
Traditional subtitles show a full line or two for a few seconds, then switch. Word-by-word captions break the text into small chunks and reveal them in rhythm with the voice. They’re usually burned in, meaning drawn into the video pixels, so they appear on every platform the same way. The burned-in captions entry explains the trade-offs, and captions vs subtitles clears up the vocabulary.
They’re related to kinetic typography, where text itself becomes the animation. The difference is purpose: captions transcribe what’s said; kinetic titles emphasize key ideas. In a good explainer, both work together: the caption carries every word, and a big animated title punches up the key word at the right moment.
Why they work
1. Sound-off viewing is common
People scroll in public transport, offices, waiting rooms and late at night. Feeds often start videos muted, or viewers keep their phone on silent. Without captions, a narrated explainer is just moving shapes. With captions, the message survives the mute button. That alone is a strong reason to caption every short video.
2. Reading along helps comprehension
Short explainers move fast: 140 to 155 words a minute at a natural pace, sometimes more. Seeing the words while hearing them gives the viewer a second chance to catch a term they missed: a number, a name, a technical word. This is especially useful for viewers watching in a language that isn’t their first.
3. They make your video accessible
Captions let deaf and hard-of-hearing viewers follow your video. That’s an audience you shouldn’t exclude, and in many contexts (education, public sector, corporate training) accessible video is expected or required. Burned-in captions help, but providing a caption file as well is better practice, since players can then display text the viewer can control.
4. Motion keeps the eye on the screen
Words appearing in rhythm give the eye something to follow between visual changes. In a video built from diagrams and titles, the caption becomes a steady thread through every scene.
How to do them right
Timing: follow the voice, not the clock
Badly timed captions are worse than none: they make the viewer read something the voice hasn’t said yet, or lag behind it. The reliable method is to derive caption timing from the audio itself, using word-level timestamps from the speech engine or a forced-alignment tool, rather than splitting the text evenly across the duration.
- Show a word when it starts being spoken, not before.
- Group words into short chunks (one to four words) that break at natural pauses.
- Never split a number, a name or a unit across chunks: “$1,000” and “8 minutes” stay together.
- Leave the last chunk of a sentence up long enough to read before the next one replaces it.
Readability: size, weight and contrast
- Large, bold, sans-serif text. On a phone, captions need to be readable at arm’s length.
- High contrast. Light text with a dark outline or shadow, or dark text on a light band. Test it over your brightest and darkest scenes.
- One highlight color, used consistently. Highlighting the current word helps, but don’t make every word a different color.
- Few words on screen. One to four words at a time, at most two short lines.
- Sentence case. ALL CAPS reads slower over many words; save it for a single emphasized word.
Placement: stay in the safe zone
On TikTok, Reels and Shorts, the bottom of the screen is covered by the post caption, the account name and the sound, and the right edge by the action buttons. Captions placed there get hidden. Keep them inside the safe zone, typically in the middle or lower-middle area, clear of the interface. Our safe zones guide shows the areas to avoid on a 1080×1920 frame.
Also avoid stacking captions on top of the key visual. If a diagram fills the center, the caption should sit just below it, not over it.
Accuracy: captions are part of the script
Automatic transcription makes mistakes with names, numbers and jargon. Proofread every caption, especially terms specific to your field. If the narration says “eighteen” and the caption shows “80”, viewers will notice. Writing a clean script first and generating the voice from it avoids most of these errors, because the caption text is already known.
Burned-in captions vs. caption files
| Burned-in (word-by-word) | Caption file (SRT) | |
|---|---|---|
| Looks the same everywhere | Yes | No, depends on the player |
| Animated, styled, highlighted | Yes | Limited |
| Viewer can turn off or resize | No | Yes, on supporting players |
| Readable as text by platforms | No | Yes |
| Easy to translate or fix later | No, needs a re-render | Yes, edit the text file |
The best approach for short explainers is both: burned-in word-by-word captions for the feed, and an SRT file to upload as closed captions where the platform supports it (YouTube does, for example). If you only have a script, the SRT caption generator builds a timed file from it, and the caption length checker flags lines too long to read.
Captions in other languages
If you publish in several languages, captions must match the narration in each one, not a translation of the English captions laid over a different voice. Word lengths change between languages (German compounds, for instance, are long), so check that chunks still fit on screen. Our AI voiceover guide and the Spanish voiceover page cover multilingual narration in more detail.
A captioning checklist
- Final script proofread, numbers and names checked.
- Captions timed from the real audio, word by word.
- One to four words per chunk, breaking at natural pauses.
- Large bold text, strong contrast tested on light and dark scenes.
- Captions inside the safe zone, not covering the main visual.
- One consistent highlight color.
- SRT file exported and uploaded where supported.
- Watched once with the sound off: can you follow the whole video?
Common mistakes
- Captions that appear before the word is spoken.
- Full sentences crammed into one line, too small to read on a phone.
- Bright text with no outline over a bright background.
- Captions sitting under the platform’s buttons and description.
- Uncorrected auto-transcription errors on key terms.
- Changing caption style from one video to the next, so your channel never looks consistent.
How Motiongist handles captions
Every Motiongist video gets word-by-word captions in the video’s language, timed from the narration itself, so each word appears as the voice says it. They’re placed inside the vertical safe zones of a 1080×1920 frame, alongside the animated titles and diagrams. You also get an SRT captions file with each video, for platforms that accept closed captions. Captions follow the video’s look; there are no separate caption style presets to pick from. Captions are available in all 10 narration languages, from English to Indonesian.
Word-by-word captions aren’t decoration. They’re how a narrated explainer works for everyone: muted viewers, second-language viewers and deaf viewers. Time them to the voice, keep them short and readable, place them safely, and ship a caption file alongside.