Most Reels are first watched with the sound off, in a feed, on a bus, or late at night. Captions are what keep those viewers watching. They also make your video accessible to deaf and hard-of-hearing viewers, and they rescue every Reel whose audio gets drowned out by music or room noise. Edits generates captions automatically from your speech, and the whole job, from generation to a clean, styled result, takes about ten minutes.
By Alex Carter, Editor · Published October 24, 2026 · Updated October 24, 2026
30-second answer: Tap Captions, pick your spoken language, and let Edits transcribe your audio. Then tap any wrong word to fix it, choose a readable style, and check that each line lands on the word it belongs to before you export.
In this guide
- Why captions matter
- Generate automatic captions
- Fix transcription mistakes
- Style captions for sound-off viewing
- Fix caption timing
- Troubleshooting
- Frequently asked questions
Why captions matter
Captions change the first three seconds of a Reel. Without them, a muted viewer sees a moving mouth and keeps scrolling. With them, the same viewer reads the hook and stays, often turning the sound on once they are invested. That is why creators who caption consistently see better retention, and retention is what the algorithm rewards. Beyond reach, captions are simply good manners: they include viewers who cannot hear your audio, and they let everyone else follow along in a noisy room.
Captions also protect your message. If a viewer watches with the sound off and your point lives only in the audio, the Reel fails for them no matter how good the visuals are. Think of captions as a second channel for the same idea, not as decoration. Get them right and the video works twice, once with sound and once without.
Generate automatic captions
Open your project and make sure the timeline is trimmed first, since captions follow the clips you have. Our trimming guide covers that step. Then tap the Captions tool, choose the language you actually spoke, and start generation. Edits listens to the audio track and writes a caption line for each spoken phrase. Short projects finish in under a minute; longer ones take a little more, and you can keep working while it runs.
When generation finishes, play the Reel once with the captions on and the sound low. You are not checking for perfection yet, just for disasters: a whole missing sentence, a wrong language, or captions attached to the wrong clip. Fix the big problems first, because small word edits are wasted on a caption track that needs regenerating. If the transcript is mostly wrong, delete it and regenerate with the correct language selected; a second pass with the right setting beats an hour of manual repair.
Fix transcription mistakes
Automatic captions get most words right and some words confidently wrong, especially names, brands, and slang. Tap any caption line to edit its text directly. Fix proper nouns first, since a misspelled name looks worse than a dropped filler word. Then read each line as a sentence: if a line runs too long, split it at a natural pause so viewers can read it in time, and if two tiny lines say almost nothing, merge them. Aim for lines a viewer can read in about two seconds, roughly seven to ten words each.
Watch for the errors machines make most: missing negatives (a lost “not” reverses your meaning), numbers heard as words, and homophones in noisy audio. A single wrong “not” is worth more correction time than ten typos, because it changes what you said. Read the captions once as pure text, without the video, and the remaining mistakes jump out.
Style captions for sound-off viewing
Readable beats fancy. Pick a bold, clean font in a size that survives a phone screen, white text with a dark outline or a solid background bar, and keep the position consistent, usually lower third, clear of the app’s own buttons and captions. If your Reel has burned-in text titles, which our text and titles guide covers, keep captions away from them so the two never overlap. Word-by-word highlighting looks lively, but use it sparingly; on a fast Reel it can feel frantic, and on a calm talking-head video it adds nothing.
Test the style on a real phone, not just in the editor preview. Export a short test clip and watch it in the feed at arm’s length. If you squint, the text is too small or too low-contrast. One style per video is a good rule; switching fonts mid-Reel looks accidental rather than creative.
Fix caption timing
A caption that appears a second early spoils the line; one that appears late leaves the viewer confused. After styling, play the Reel and watch each line’s entrance against the spoken word. If a line is early or late, drag its edges on the timeline or nudge it in the caption editor until it matches. Split long lines at pauses rather than letting one block cover two sentences, and never let a caption linger after the speaker has moved on, since a stale line reads like an error.
Troubleshooting
- Captions are in the wrong language. Delete the track and regenerate with the correct spoken language selected before you start.
- Whole sentences are missing. Quiet speech and heavy music bury words. Raise the voice track or regenerate from a cleaner audio mix; our advanced audio guide shows how.
- Captions sit under the app buttons. Move them higher in the safe zone and re-export a test clip to confirm.
- Style changes do not stick. Apply the style to the caption track, not a single line, then check the next line before exporting.
- Captions vanish in the export. Make sure the caption track is enabled and visible in the project, then export again at the settings in our export settings guide.
Caption habits that pay off
Caption every Reel, even the short ones. A five-second video with no speech still benefits from a caption that states the point, because the viewer deciding whether to stay reads before they listen. Keep a consistent caption style across your account so viewers recognize your videos at a glance; the style becomes part of your brand. And caption before you add music, not after: music makes it harder to hear the words you are checking, and a caption pass done over a finished mix catches fewer mistakes.
One more habit: export a captioned test version for yourself before the final export. Watch it once in the feed, muted, from a normal scrolling distance. If you can follow the whole idea without sound, the captions did their job. If you cannot, the fix is almost always shorter lines or larger text, not a different font.
Get the official app
Captions are free in the official app, and your next Reel is the perfect test. Get Edits from the official stores below, generate captions on a real clip, and fix every wrong word before you post. Next: Text and Titles in Edits →
Frequently asked questions
Are captions in Edits automatic?
Yes. Edits can generate captions from your audio automatically, in the language you choose, and you can edit every word afterwards.
Do captions cost extra?
No. Caption generation is part of the free Edits app, and exports keep the captions you styled.
Why do my captions show the wrong words?
Usually background noise or a strong accent confuses the transcription. Re-record in a quieter room, speak slightly slower, or correct the words by hand afterwards.
Can I style captions myself?
Yes. You can change the font, size, color, and position, and highlight words as they are spoken.
Do captions help people find my Reel?
They help retention, which helps reach, and they make your Reel watchable with the sound off, where most browsing happens.
Can I add captions on a PC or Mac?
Yes, inside an Android emulator running the same free app. Our PC guide and Mac guide explain the setup.
Edits for PC is an independent, unofficial guide. It is not affiliated with Meta Platforms, Inc. or Instagram, and it does not host or sell the Edits app.
Captions for different video types
Talking-head videos want complete, accurate captions, because the words are the content. Do not summarize or paraphrase; viewers who read along will notice the mismatch, and it undermines trust. Tutorial videos want captions that match the steps: keep each caption on the action it describes, and add a short caption for on-screen actions that have no speech, like “tap Export,” so a muted viewer can still follow the lesson. Comedy wants captions timed to the punchline, not ahead of it; a caption that arrives early ruins the joke the same way a spoiler does.
Music-led and B-roll videos need the least captioning, but they still need a line or two stating the point, because a muted viewer otherwise has no reason to stay. Product videos sit in the middle: caption the benefit, the price if you mention one, and any instruction, and let the visuals carry the rest. The rule across all types is the same: captions should carry exactly the information a muted viewer needs, no more and no less.
When captions hurt
Captions help until they cover the thing viewers came to see. A caption block over a speaker’s face, over a product close-up, or over burned-in text you already added turns help into clutter. Move captions to clear space, shorten the line, or drop captions from shots where the picture carries the whole message. Likewise, do not caption song lyrics word for word unless the lyrics are the point; a wall of lyric text distracts from a visual sequence. Captions should serve the video, and on some shots the best caption is none at all.
The test is simple: watch the Reel once with captions on and once with them off. If the captioned version hides anything important, move or cut the captions on those shots. Captions are a layer, not a blanket.
One last check before you export: watch the first five seconds muted. If the captions there hook you, the rest will follow. Most viewers decide in those first seconds, and captions are the cheapest way to win them.