How to add captions to Reels (word-level, burned in)

Most Reels play with the sound off. If the first sentence is not on screen, the viewer is already gone. Instagram can auto-caption a post, but those captions live inside the app: they lag, they ignore your type style, and they vanish when someone saves the file. Burned-in captions are pixels in the MP4. The same file works on Instagram Reels, TikTok, and YouTube Shorts.

ClipoStack is built for talking-head footage. You upload a clip. Whisper transcribes it with a timestamp on every word. Captions lock to the spoken beat instead of sitting on a guessed delay. You review the text, pick a readable style, and export one vertical file with the captions already in the frame.

This guide is the captions job only. TikTok and Shorts are covered below as extra sections, not extra URLs. You do not need a separate how-to for TikTok captions. If you also want cutaways on the same export, follow the B-roll guide after this flow.

Six steps in the studio

  1. Upload your clip — drop an MP4, MOV, MKV, AVI, or WebM into the studio. Horizontal talking-head is fine; the export is a vertical short. Transcription starts automatically.
  2. Wait for word-level transcription — speech becomes text with a timestamp on each word, so captions can appear on the beat instead of a half-second late. ClipoStack uses Whisper-large-v3-turbo. Most European languages work without extra setup.
  3. Review captions locked to speech — when the transcript is ready, captions sit on the timeline against the audio. Each word can highlight as it is spoken. If a name or brand is wrong, you fix the text before you render. Placement and how many lines stay on screen are editable. Nothing in the caption track is locked.
  4. Render burned-in captions — export an MP4 with the captions rendered into the frame. That is the file you post. You are not attaching an SRT that Instagram might ignore, and you are not hoping in-app auto-captions match your cut.
  5. Post to Reels — upload the file to Instagram Reels. Because the text is already in the video, the caption look stays the same in-feed, in-profile, and if someone downloads the clip. If Instagram offers its own auto-captions on top, turn those off so two subtitle tracks do not stack.

Captions for TikTok

TikTok auto-captions are a safety net, not a look. If you care about word-level timing and a consistent style, burn the text in before you post. Export once in ClipoStack, then upload that same MP4 to TikTok. You do not need a second render or a separate captions-for-TikTok page. Keep captions out of the very bottom UI chrome so TikTok buttons do not cover the last line.

Captions for YouTube Shorts

YouTube Shorts can show YouTube’s own captions if auto-captions stay on. For talking-head shorts that should look the same as your Reels, use the burned-in export. One file, three apps. If you also add a YouTube transcript for accessibility, that is extra — the on-screen look still comes from the pixels.

Why burned-in captions beat in-app auto captions

In-app captions are convenient and often late: missing words, stacked under UI, gone after a download. Burned-in captions survive a save, a stitch, and a cross-post. Word-level timestamps keep the punchline on the spoken beat, which is the difference between a caption people read and a caption they skip.

Tips for captions that get read

  • Fix proper nouns before you render. Whisper is strong on speech, weaker on names and brands. One edit in the caption track is cheaper than a bad line in three apps.
  • Contrast over decoration. White fill with a dark outline still wins on a phone. Karaoke highlight is optional; readability is not.
  • Export once, post three times. The burned-in MP4 is the artifact. Reels, TikTok, and Shorts should get the same file so timing and type do not drift.

Common questions

Do captions work on TikTok and Shorts?

Yes. Burned-in captions travel with the file, so one export works on Reels, TikTok, and YouTube Shorts. You do not need a separate TikTok captions page or a second render.

Are captions word-level?

Yes. Whisper timestamps keep each word on the spoken beat, so captions lock to speech instead of a guessed delay.

Can I edit the text?

Caption text, placement, and style stay editable before you render. The transcript is a starting point, not a locked result.

Related: How it works for the full pipeline, or Add B-roll guide if you want cutaways on the same export.