How many words a caption should show at once

Scrolling through a vertical feed, people read each caption in one glance. They haven't committed to the video yet, the image is in motion and the sound might be off, so a caption has a single look to get its point across. A short phrase can do that. A line spanning the frame pulls the reader's eye across to the edge and back, and by then the next caption may be up.

Someone who has sat down for an interview or a twenty-minute tutorial reads differently. They are in no hurry, and a caption that flips every second wears on them. Longer lines, more like film subtitles, suit them better: the text changes less often, and every caption completes a thought.

Captionwale's templates come in three families. Which family a template is in decides how the transcript gets cut into captions. In the template list the Dynamic collection has a heading to itself, yet each of its designs is still one of the three:

A Single line template sets the number of words per caption. Most built-in designs use three; a design that doesn't name a number gets 4. Three things cut a caption short. One is a silence of more than 0.28 seconds. Another is a word carrying end-of-sentence punctuation: ., ? or !, or the Hindi full stop, the danda (।). The third is when one more word would take the row over 24 characters, with spaces counted. Captions with more than one row allow 24 characters on each row.

Whenever you change a setting the rules simply run again; nothing is transcribed afresh. The controls live in the Captions panel, under Caption settings:

When a caption stops mid-phrase, the viewer is left waiting for the rest. When one opens with the leftover end of the last phrase, it looks like a mistake. People usually take a breath between ideas, so pauses take care of most breaks. If someone talks without stopping, the limits on words and characters decide instead, and they may slice a phrase in two.

Hindi and Hinglish put their small words after the word they belong with. Postpositions come after the noun, and helper verbs after the main verb. Separate 'mein' from 'Surat', or 'raha hai' from 'kar', and the leftover piece is meaningless on its own. In English the order flips: 'the' and 'a' belong with the word after them.

To repair a break, click the word that ought to open the next caption, then press Enter or choose 'Split line here'. Use 'next line →' or '← previous line' to push one word over a break, which also rescues a word that a pause has cut away from its phrase. Any of these can be undone.

Across the Indian scripts, Devanagari, Gujarati and Tamil included, a vowel sign attaches to its consonant from above, below or the side. Swap one and the word changes: दीन means poor, while दिन means day. Because those signs are among the finest strokes in any caption, a size chosen for English can leave them unreadable on a phone. Make the text bigger in the 'Size & position' pane.

The signs add height as well. Stacked rows of an Indian script therefore need more breathing room than English rows, or the marks on one row bump into the letters of the next. 'Line spacing', found in the Spacing pane of Multi line templates, opens up that gap.

In Tamil, Telugu, Kannada and Malayalam, a root collects ending after ending, and one spoken word can fill most of a row by itself. Captions in these languages therefore get fewer words from the 24-character rule automatically. Since it counts characters and not the space they take on screen, check how it looks in the preview. Which fonts draw each script cleanly is covered in picking caption fonts for Indian scripts.

You won't find a reading-speed figure here, because no single number suits all languages, scripts, speakers and audiences. Treat Captionwale's defaults as a starting point; they don't judge your particular video.

Only timing and punctuation reach the grouping. It has no idea that a break falls inside a name, cuts a number away from its unit or leaves a helper verb stranded, and sparse punctuation in the transcript means fewer breaks at sentence ends. Any silence over 0.28 seconds starts a fresh caption, even mid-phrase when the speaker slows down for effect, and there's no setting to change that. Only '← previous line' can carry a word back across, and only one word per press. Before export, scan the caption list from top to bottom; afterwards, watch the exported file on a phone held at viewing size.