How to add captions to TikTok, Reels, and Shorts

Around 80% of short-form video is watched with the sound off. If your first three seconds do not read on a muted phone screen, most viewers scroll past before they know what the clip is about. Captions are the fix, and every serious creator on TikTok, Reels, and Shorts is already using them.

Why captions decide whether a Short gets watched

Muted viewing is the default on all three platforms. Reported figures for how many people watch short-form on mute sit around 80%, and platform data has shown captioned TikToks lifting impressions by about 55% versus uncaptioned uploads. The number that matters most is retention in the first three seconds, because that is where the algorithm decides how far to push the clip. A caption gives someone a reason to keep watching before the audio ever loads.

There is a second reason: search. TikTok, Reels, and Shorts all index caption text for in-app search, and Google surfaces short-form video in results. Captions are the transcript that lets that indexing happen.

The three ways to caption a Short

Captions on short-form video split into three families. They cover different tradeoffs, and picking the wrong family is the reason a lot of creators end up re-doing the work.

Platform-native captions

TikTok, Reels, and Shorts each have a built-in auto-caption button.

TikTok: after you pick your clip, tap Captions in the right toolbar. Auto-transcribes, lets you edit lines and pick a style.
Instagram Reels: add the Captions sticker during the editing step. Instagram transcribes, you tweak.
YouTube Shorts: the caption toggle in the Shorts editor auto-generates lines and lets you switch style.

The native flow is free, fast, and integrated with the upload. The tradeoff is that the captions only exist on that upload. If you want to cross-post the same clip to another platform, the captions do not carry. If you want the video file with captions burned into the pixels, so you can reuse it in a compilation, a website, or an ad, the native flow does not give you that.

Online caption generators

A large field of web tools take a video upload and return a captioned MP4: CapCut, VEED, Kapwing, Submagic, Descript, and more. Most auto-transcribe, offer preset styles, and export a hardcoded file you can post anywhere.

The tradeoff here is different. You upload the raw video to a server you do not control. Free tiers often add a watermark or cap the export length. Some cap you at 720p until you pay. And most tools are opinionated about styling: you pick a preset and nudge it, but the caption engine underneath is closed.

A browser-based editor

The third shape is what tscaps is: a browser editor that runs the caption pipeline on the page you are already looking at. You drop the video into the tab, it transcribes with word-level timings, you style the captions in a UI or in raw CSS, and it exports an MP4 with the captions burned in. Nothing gets uploaded on the local surface; the video file stays on your machine the whole time.

The tradeoff is up front, not hidden. Local transcription runs a Whisper model in a Web Worker on your device, so a fast laptop gets a fast result and a low-end phone gets a slow one. That is why the cloud lane exists too: server-side transcription (ElevenLabs Scribe v2) runs faster and handles noisy audio and mixed languages better, and it enables the AI styling passes (scene-aware styling, multi-speaker detection, word emphasis, contextual emojis). The cloud free tier adds a watermark on exports; higher tiers lift the watermark and the per-period cap.

How to burn styled captions into a TikTok, Reel, or Short with tscaps

The flow is the same across TikTok, Reels, and Shorts because the output is a vertical MP4 you can post to any of them. Four steps:

Drop the video. Open tscaps.io, drag your file in. Nothing uploads on the local lane; on the cloud lane the file goes to the transcription service and then to the editor.
Wait for the transcription. You get word-level timings back. Word-level, not sentence-level, is what lets each word appear in sync with the audio.
Pick a template and edit. Templates are CSS plus a bit of JSON, and there is a gallery to start from. You can swap words, split segments, and change styling per word if you want.
Export. The engine rasterizes the caption DOM at every frame and composites it with the video, in the browser, through MediaBunny. What you see in the preview is what ships in the MP4.

That MP4 is a normal file. Post it to TikTok, Reels, and Shorts without re-captioning each one.

Caption design that survives a 6-inch screen

The reason platform-native captions all end up looking similar is that the constraints are physical. You have around six inches of screen and a viewer who is not paying full attention. A few things that hold up across TikTok, Reels, and Shorts:

Font: a bold sans-serif is the safe default. Impact, Bebas Neue, Montserrat Bold, Inter Bold. Serifs and thin weights fall apart at low resolution.
Size: aim for caption text that fills roughly 6 to 10% of the video height. Too small and it reads as a subtitle for a movie you did not want to watch. Too large and it hides your subject.
Word count: one to five words per line. Word-by-word animation, where each word pops as it is spoken, tracks the fast pace of short-form better than full-sentence blocks.
Placement: center-lower is the neutral choice, but leave the bottom 15% clear so your caption does not collide with the TikTok username bar, the Reels action buttons, or the Shorts UI.
Contrast: white text with a hard drop shadow or a solid stroke reads over almost anything. A translucent black background block is the fallback when the video is chaotic.

Most tools give you these as presets. tscaps ships a template gallery with the same defaults, and every template is a CSS file you can open and edit if you want a specific look.

Cloud or local: which lane fits you

The two tscaps surfaces exist for different readers, not for different price tiers.

Local is at tscaps.io/local. No upload, no account, no watermark on the free path. Best when your audio is clean, you are on a laptop or a decent desktop, and you care that the video does not leave your machine.
Cloud is at tscaps.io. Server-side transcription that is faster and handles noise better, plus the AI styling passes. Every cloud tier gets the AI features, including anonymous use. The free path adds a watermark to exports; paid tiers lift the watermark, raise the per-period cap, extend the max clip length, and sync projects across devices.

If your audio is clean and you are on a real computer, local is usually the better call. If you are on a phone, the audio is noisy or multi-speaker, or you want AI to do the styling for you, cloud is faster and gives you more.

FAQ

Do I need to add captions if TikTok already generates them?

Native captions are fine if you post exclusively to TikTok and always with the native editor. Once you cross-post to Reels or Shorts, or you export the MP4 for anywhere else, native captions do not carry. Burned-in captions do.

What is the difference between burned-in and platform captions?

Burned-in captions are pixels in the video frame. They cannot be turned off, and they travel with the file to any platform. Platform captions are metadata the app draws on top at playback time, and they only exist in that app.

What font size and style works for TikTok, Reels, and Shorts?

Bold sans-serif, one to five words per line, caption height around 6 to 10% of the video, white text with a drop shadow or a stroke, centered and above the bottom 15% so the platform UI does not overlap it.

Can I add captions without uploading my video?

Yes on the local lane. The Whisper model runs on your device in a Web Worker and the export runs on your device too, so the video never leaves the browser. Cloud transcription needs the audio on the server for the ElevenLabs pass.

Ready to caption a Short? Drop a clip at tscaps.io.