How to Add Captions to Reels, Shorts and TikTok
Every short video needs two caption layers and only one of them survives. Here is what each platform actually accepts, and which layer is your job.

There are two kinds of captions on a short video, and mixing them up is why your text vanishes on somebody else's phone. Burned-in captions are pixels inside the video file and cannot be switched off. Platform captions are a separate track the viewer controls. You want both, and only one of them is your job in the edit.
Which one each platform gives you, and what it accepts from you, is below. Every line is from the platform's own documentation.
What does each platform do with captions?
Instagram Reels
- Automatic closed captions, generated by speech recognition
- Instagram's help page states the speech is written out as text at the bottom of the reel
- No documented route for uploading your own caption file to a reel
- Placement is not yours to choose
Source: Instagram Help Center, Manage closed captions for your reels on Instagram.
TikTok
- Auto captions are selected on the editing page after you record or upload
- You can edit the generated text before you post
- TikTok states that captions added by creators are part of the video content, and that a viewer can turn them off from the share panel
- Announced 6 April 2021, initially in American English and Japanese
Source: TikTok Newsroom, Introducing auto captions, 6 April 2021.
YouTube, including Shorts
- Accepts uploaded caption files. Basic supported formats include SubRip (.srt) and SubViewer (.sbv)
- YouTube's documentation says only basic versions of these files work, no style markup is recognised, and the file must be plain UTF-8
- Scenarist Closed Caption (.scc) is YouTube's stated preferred format, which matters for broadcast work and almost never for a Short
- Automatic captioning runs as well
Source: YouTube Help, Supported subtitle and closed caption files.
- Documents adding closed captions to a video post by uploading an SRT file, alongside auto captions in the composer
- Captions can be edited after the post goes live
Source: LinkedIn Help, Add closed captions to videos on LinkedIn.
Why platform captions are not a retention tool
Three reasons, and the third is the one that costs you views.
- The viewer can switch them off. TikTok documents the exact path: share panel, captions button, off. A caption layer a viewer can remove is not a layer you can build a hook on
- They are generated after you upload, so they play no part in how you paced the cut. You cannot time a line to land on a beat that does not exist yet
- They render where the platform puts them. Instagram's own wording is that the text goes at the bottom of the reel, which is the same region the interface furniture occupies. Text placed there competes with the caption field, the buttons and the progress bar
Accessibility and retention are two different jobs. Platform captions do the first well and the second not at all. That is not a criticism of them, it is what they were built for. TikTok introduced auto captions explicitly as an accessibility feature for people who are deaf or hard of hearing.
Burned in or uploaded: which one, when
- Burn in the words you need read. The hook line, any number, any name, the punchline. Those are load bearing and cannot be optional
- Upload a file for the full transcript where the platform takes one, which today means YouTube and LinkedIn. It costs you two minutes and it is the accessible version
- Never let auto captions carry a clip whose meaning depends on one specific word landing correctly. Speech recognition on a product name or a surname is where this fails most often
- If a clip goes to four platforms, burn in once and export once. Four hand-placed caption passes is four chances to get it wrong
What breaks a caption after you export
- Styling gets stripped. Nothing you style inside an SRT survives, because YouTube reads only basic versions of the file and recognises no markup. Style lives in the burn-in, never in the file
- Encoding. YouTube requires plain UTF-8. A file saved in another encoding comes back with mangled apostrophes and accented characters, and it usually goes unnoticed until someone reports it
- Placement. Burned-in text sitting low in a vertical frame gets covered by interface furniture that is not there in your editor, which is a preview problem, not a design problem
- Re-encoding. Every platform recompresses what you upload. Thin strokes and low contrast are the first things to soften, so a caption set in a light weight at a small size is the one that becomes hard to read. Heavy weight, hard contrast, generous size
A caption pass that takes four minutes
- Transcribe automatically in your editor. Do not type it out
- Fix the proper nouns first. Product names, people, places, anything with a number in it. This is where machine transcription reliably misses
- Set the caption box above the bottom interface zone rather than centred at the bottom out of habit
- Cap each line at about four words for vertical. Two short lines read faster than one wide one at arm's length
- Burn in, then export once to your master settings so the text quality survives the platform's re-encode
- Upload the SRT as well on YouTube and LinkedIn. Same transcript, thirty seconds of work, and it is the version a screen reader user gets
Key takeaways
- Burned-in captions are pixels and permanent. Platform captions are a track the viewer can switch off
- Instagram places automatic captions at the bottom of the reel, which collides with the interface zone
- TikTok generates captions on the editing page, lets you edit the text, and lets viewers disable them
- YouTube accepts SRT and SBV, recognises no style markup, and requires plain UTF-8
- LinkedIn accepts an SRT upload alongside its auto captions
- Burn in the load-bearing words, upload a file for the full transcript, and never rely on speech recognition for a name
- Heavy weight and hard contrast survive platform re-encoding. Thin and small do not
Frequently asked questions
Do burned-in captions hurt reach?
There is no platform documentation saying they do. What is documented is that all four platforms offer captions in some form and treat them as an accessibility feature. Burning in your key lines and also supplying a file where accepted covers both audiences without any known penalty.
Can I upload an SRT file to Instagram Reels?
Instagram's help pages document automatic closed captions for reels rather than a file upload route. If you need caption text you control on Instagram, burn it into the video during the edit. That is the only placement and styling you fully own there.
Why do my captions look blurry after posting?
Every platform re-encodes your upload, and fine detail suffers first. Thin fonts, small sizes and low contrast against busy footage are the usual causes. Increase weight and size, add a solid or heavily shadowed backing, and export at your platform's full supported resolution.
Does YouTube accept SRT files on Shorts?
YouTube's supported caption file documentation covers uploads to your videos and lists SubRip among the basic formats. Style markup is not recognised and the file must be plain UTF-8, so treat the uploaded file as the transcript and keep all visual styling in the burn-in.
If the caption pass is the thing you keep skipping
Captions are four minutes a clip until you are shipping fifteen clips a week, at which point they are an hour nobody has. That is usually the point at which founders start pricing this out, and it is worth knowing what short form editing actually costs before you do. EchoPulse does this work for founders and business owners as part of a full pipeline.


