How many words to put on screen at once
By AutoCap · 31 October 2026
One word, a short phrase or stacked rows: how to size captions for sound-off viewing, where to break lines, and what AutoCap's caption grouping does by default.
Glances and settled reading
On a vertical feed, a caption is read in a glance. The viewer is still deciding whether to stay, the picture is moving and the sound may be off, so each caption has to land in one look. A short phrase can. A line across the whole frame sends the eye to the far edge and back, and the caption may change before it returns.
Someone who chose a twenty-minute tutorial or an interview reads differently. They are settled, and captions that change every second become tiring. Fuller lines, closer to film subtitles, suit them: fewer changes, each caption a complete thought.
One word, a phrase or stacked rows
Every AutoCap template belongs to one of three families, and the family decides how the transcript is cut into captions. The template list gives the Dynamic collection a heading of its own, but each design in it is one of the three:
- One word: each word appears alone as it is spoken. It gives hooks and punchlines a strong rhythm, but the viewer has to assemble each sentence from memory, so it suits short lines better than explanations.
- Single line: a short phrase on one row. It is the general-purpose choice for Reels and Shorts, and the only family where you set the number of words yourself.
- Multi line: rows stacked and composed as one design, some arranged around a single highlighted word. It holds more words without any row getting wider, which helps with slower speech and phrases meant to be read as a unit.
What AutoCap does unless you change it
On a Single line template a caption holds as many words as the template was designed for: 3 on most of the built-in designs, and 4 on any that names no number. It ends sooner in three situations: the speaker pauses for more than 0.28 seconds; a word carries sentence-ending punctuation, whether ., ? and ! or the danda (।) that closes a Hindi sentence; or the next word would take the row past 24 characters, spaces included. A caption with more than one row gets 24 characters for each row.
These rules rerun whenever you change a setting, with no new transcription. The controls sit under Caption settings in the Captions panel:
- 'Max words per caption' goes from one to eight. It appears only on Single line templates, because One word and Multi line designs set the count themselves.
- 'Min duration' keeps a short caption on screen for longer, up to five seconds, without merging it into the next one.
- On a Multi line template built row by row, the Rows pane lets you restyle one row on its own.
Break where the speaker breaks
A caption that stops halfway through a phrase leaves the viewer waiting, and one that opens with the tail of the last phrase reads like a typo. Pauses do most of the work, since speakers tend to breathe between ideas; when someone talks without stopping, the word and character limits decide instead and can cut a phrase in two.
In Hindi and Hinglish the small words come after the word they belong to: postpositions follow the noun and helper verbs follow the main verb. A break between 'Surat' and 'mein', or between 'kar' and 'raha hai', leaves a fragment that means nothing alone. English runs the other way, so keep 'the' and 'a' with the word after them.
To mend a break, select the word that should begin the new caption and press Enter, or use 'Split line here'. The '← previous line' and 'next line →' buttons move one word across a break, including a word a pause has cut off from its phrase, and every change can be undone.
Why Indian scripts need more room
In Devanagari, Gujarati, Tamil and the other Indian scripts, vowel signs attach above, below or beside the consonant, and they change the word: दिन (day) and दीन (poor) differ by one vowel sign. Those signs are among the smallest strokes in a caption, so a size that suits English can leave them hard to make out on a phone; raise it in the 'Size & position' pane.
Those marks also take up height, so stacked rows in an Indian script need more space between them than English does, or one row's signs crowd the next row's letters. On Multi line templates, 'Line spacing' in the Spacing pane opens that gap.
Tamil, Telugu, Kannada and Malayalam build long words by adding endings to a root, so one spoken word can fill much of a row. The 24-character limit gives those captions fewer words on its own, but it counts characters rather than on-screen width, so judge the result in the preview. Choosing fonts for Indian-language captions covers which typefaces draw each script cleanly.
What the grouping can't judge
This guide gives no reading-speed figure, because one number can't serve every language, script, speaker and audience. AutoCap's defaults are a starting point, not a verdict on your video.
The grouping sees only timing and punctuation. It can't tell that a break splits a name, separates a number from its unit or strands a helper verb, and a transcript with little punctuation gets few sentence-end breaks. A pause over 0.28 seconds starts a new caption even inside a phrase slowed for effect, and no setting changes that; only '← previous line' pulls a word back across it, one word at a time. Read down the caption list before exporting, then watch the file on a phone at the size your viewers will.
Questions
How many words should a caption on a Reel have?
There is no single right number. Most of AutoCap's Single line templates start at three, which can usually be read at a glance; use two for rapid speech, or four or more for a slower video.
Why did my captions get shorter when I changed template?
Each template brings its own word count, and choosing one applies it. Pick the design first, then set 'Max words per caption' if it is a Single line template.
Can I change how long a pause has to be before a new caption starts?
No, it is fixed at 0.28 seconds. To keep a word with the caption before it, select the word and use '← previous line'.
Do word-by-word captions work for long explanations?
They tend not to, because the sentence is never on screen as a whole. A phrase or stacked template is easier to follow there, and I'd keep one word at a time for hooks and punchlines.