The right number of words

Why caption length matters

On a phone, people read the captions while watching the person talking. Fewer words per caption means less to read each time but more changes per second, so the right length depends on how fast you speak and how much is happening in the shot.

Longer captions also raise the stakes on where each one ends. A five-word caption that stops mid-phrase is harder to follow than a two-word one.

Setting caption length in the studio

1. Add your clip, run Find words and pick a preset with roughly the length you want.
2. Open the Layout tab and, under Caption length, choose One word or Phrases.
3. For Phrases, set Most words per caption (2 to 10) and Most letters per caption (8 to 60), and choose One line or Up to two.
4. Use Start a new caption after a pause of so a caption ends when you stop talking, and Keep the last word on screen for so it lingers a moment after.
5. Play it back at full speed and adjust until you can read every caption once without pausing.

Captions too fast to read?

Captions that flash past too fast, a common complaint about CapCut's auto captions, hold too few words for how quickly you talk. A caption stays up as long as it takes to say, so more words per caption means each one stays on screen longer.

• Switch Caption length from One word to Phrases, or raise Most words per caption by one or two.
• Raise Keep the last word on screen for, so the last caption of each sentence lingers.
• If captions break at every short breath, raise Start a new caption after a pause of.
• Check at full speed with the sound off: if you can read each caption once, it is slow enough.

Three common lengths

• One word: big, punchy and hard to skim past. It suits hooks, fast talkers and edits with lots of cuts. Clean, Tall and Glow work this way.
• Two to four words on one line: the usual choice for talking clips. Comic shows two words, Bold and Rounded up to three, Boxed up to four.
• Five or six words on up to two lines: easier for slower speech, interviews and explainers, where a whole phrase helps more than speed. Karaoke, Typewriter and Editorial work this way.

A word limit alone treats 'a' and 'unbelievable' alike, which is why Most letters per caption exists. For English transcripts, a phrase model then chooses where inside those limits each caption ends.

Questions