Before you start: clean audio makes better captions
Auto captions are only as good as the audio they listen to. Wind, music playing over your voice and a distant microphone all cause missed or wrong words, and every wrong word is one more correction later. Before you generate captions, lower any background music, make sure your voice is clearly audible, and trim off any silence at the start. If you need to tidy the clips first, follow our trim, split and reorder tutorial.
Menu names can vary slightly between app versions, so the steps below describe the controls by what they do. If something looks different on your phone, look for the Captions or Text tools in the editor toolbar.
Add auto captions
- Open your project in Edits and make sure the clip with speech is on the timeline.
- Open the Captions tool (or the Text tools, depending on your version) from the toolbar.
- Choose automatic captions and confirm the language you speak in the video. Picking the wrong language is the most common reason for gibberish.
- Generate the first caption track and wait while Edits listens and writes the words.
- Play it back from the start. Don't export yet: this is the draft you are about to correct.
Correct the words
Tap any caption line to edit its text. The fastest method is a single pass from start to finish, fixing as you go. The usual mistakes are:
- Names and brands, which speech recognition guesses at, such as a misspelled product name.
- Sound-alike words like their, there and they’re, or to, too and two.
- Numbers and prices, which can come out as words or the wrong digits.
- Slang and fast speech, where several words merge into one.
Read each line aloud against the audio as you play it. If you find the same wrong word repeatedly, correct each instance; every line is its own piece of text.
Fix timing and line breaks
Right words on the wrong beat still feel wrong. Each caption sits on the timeline as a block, and you can adjust when it appears and disappears so it lands with the speech. Two quick rules:
- Keep lines short. Two short lines read faster than one long one. Break a long caption at a natural pause rather than mid-phrase.
- Match the voice. A caption should appear just as the words start and leave soon after they finish. Don't let the text lag behind the speaker.
If a caption block is too long, shorten the text or split it into two; the same split skill from the trim and split tutorial applies to your timeline.
Style your captions so they stay readable
You can change the look of captions, including size, colour and placement. Readability beats decoration:
- Contrast first. White text over a dark outline or box reads on almost any footage. Thin pale text on a bright background disappears.
- Mind the safe area. Instagram puts its buttons and your username over the bottom and right of a Reel. Keep captions above the very bottom so nothing covers them.
- Stay consistent. Use one style across a whole video. Highlighting the spoken word can add energy, but pick one effect and keep it.
For titles and other on-screen text, see the text and titles tutorial.
Add a second language with bilingual captions
Edits can add a second caption track in another language alongside the original, so one Reel shows text in two languages. After your first track is corrected, choose the translation option (it may be labelled Add translation, depending on version), pick the second language, and generate the translated track. Reports list 15 launch languages: English, Hindi, Spanish, Portuguese, French, German, Italian, Korean, Japanese, Thai, Indonesian, Russian, Bengali, Gujarati and Kannada. Availability can vary by region and app version, so check the picker in your project.
Two things to know: the captions are burned into the video, and the translation covers only the on-screen captions, not your spoken audio or your Instagram post caption. Translation is also machine-made and can read literally, so correct your original first, then have a reader of the second language check the translated track if you can.
Common caption problems
| Problem | Likely cause | Fix |
|---|---|---|
| Garbled or missing words | Background noise or loud music | Lower the music, improve the audio, regenerate |
| Gibberish throughout | Wrong spoken language selected | Confirm the language, then regenerate |
| Names misspelled | Recognition guesses unfamiliar words | Correct each instance by hand |
| Captions hidden behind Instagram buttons | Placed too low or too far right | Move them up into the safe area |
| Translation sounds odd | Machine translation is literal | Review the second track with a native reader |
| No captions option or nothing generates | Version, region or audio issue | See captions not working |
Audio trouble? Try audio problems, and lift a quiet voice first with the music and audio tutorial or voiceover guide. When you're done, set your quality in the export settings guide.
Edits captions FAQ
Open the Captions or Text tools in your project, choose automatic captions, confirm the spoken language and generate the caption track. Then review and correct it.
Tap the caption line and edit the text. You can also adjust its timing, line breaks and appearance.
No. Treat them as a draft. Names, sound-alike words, numbers and fast speech are the usual errors, so read the whole track before exporting.
Yes. Bilingual captions add a second caption track in another language. Reports list 15 launch languages, and availability can vary by region and version.
No. Translation applies only to the on-screen captions burned into the video, not the spoken audio or your written Instagram caption.
Yes. Edits is free, and exports have no watermark. See the pricing page.