How to Add TikTok-Style Captions to Any Video (AI Auto-Captions)

You can add captions to a video in your browser, styled like the ones all over TikTok and Shorts, where each word lights up as it's spoken. ClipCraft's Video Captions tool has 82 animated caption styles, an AI auto-caption option that writes the transcript with exact word timing, and a free manual path where you paste your script and tap along to sync it. The captions get burned into the video on export, so they play everywhere, no player support needed.

Captions matter more than most edits. A huge share of short-form video gets watched on mute, and animated word-by-word captions are the difference between someone reading along and someone scrolling past. Here's the whole workflow, including what's free and what costs tokens.

Pick a caption style first

Open Tools → Video Captionsand scroll to the style gallery. Every tile is a tiny canvas animating a sample line at 30fps, with the "spoken" word cycling through, so you see exactly how each style moves before you commit. There are 82 presets: the classic creator looks (Beast Mode, the yellow Hormozi pill, karaoke sweeps), a gothic and horror pack, pixel and gamer fonts, handwritten scripts, neon signs, and a proper sing-along Bouncing Ball that hops between words.

The search box above the grid matches vibes, so typing "horror" or "gamer" filters the wall down, and there's a row of 15 tag chips (karaoke, neon, retro, comic, handwritten, pixel...) if you'd rather click than type. Every chip returns at least ten styles.

Two ways to add captions to your video

Drop your video onto the stage (or click Choose video), then get your words in. There are two paths, and they end in the same editable caption rows.

AI auto-captions (the fast way)

The ✨ Auto captions button listens to your video with Whisper, a speech recognition model, and writes every line with per-word timestamps. That last part is the point: the highlight lands on each word at the exact moment it's spoken, which is what makes the effect look professional instead of approximate. The lines come back already split into short TikTok-length chunks.

It costs 5 tokens per started minute, and the button shows the estimate for your video before you press anything. A 2:30 clip counts as three minutes, so 15 tokens. If the job fails to start, the tokens come back automatically. One privacy detail I like: the tool extracts the audio in your browser and uploads only that. The video itself never leaves your machine.

Paste a transcript and tap to sync (the free way)

If you already have a script, paste it into the captions box. Line breaks become one caption each, and a single paragraph auto-splits into short chunks (4 words per caption by default; you can set 1 to 10, then hit Re-split anytime). Timing gets distributed across the video weighted by word count, which is a rough first guess you'll want to fix.

That's what Tap to sync is for. Press it and the video plays from the start; every tap (the button, Space, or Enter) stamps the next line's start time and closes the previous one, and a final tap ends the last caption. Esc bails out. It takes one real-time pass through your video and the result is surprisingly tight. Manually typed captions time the words within each line by length rather than by actual speech, so the AI path still wins on precision, but for a 30-second clip the tap method is honestly good enough.

Make the style yours

The Customize panel (the sideways tab next to your captions) exposes everything the presets are built from: font, size, vertical position, text and highlight colors, the highlight mode itself (word changes color, word gets a pill, karaoke fill sweep, underline, or the bouncing ball), word pop scale, outline width, glow, and uppercase/bold/background toggles. Change anything and a Reset button appears so you can get back to the preset you started from.

Exporting: what to expect

Export renders in your browser. A hidden canvas redraws every frame with the captions painted on, and the same renderer that drew your preview draws the export, so what you saw is what ships. Nothing uploads at this step.

• The export plays your video once in real time while recording, so a 60-second video takes about a minute to export. Plan for that on long clips.
• Output is MP4 where the browser supports encoding it; Chrome usually produces a WebM instead. Both upload fine to TikTok, YouTube, and Instagram.
• Resolution is kept up to 1920 on the long side.
• Each export costs 1 tool use. The free plan includes 1 tool use per month, so you get one export free; paid plans raise that (see pricing).

If you'd rather keep editing than export, the Send to VideoCraft button renders just the captions onto a transparent layer and opens VideoCraft, our free browser video editor, with your video on one track and the captions on their own layer above it. That hand-off is free, and it means you can keep cutting the video while the captions ride along as a clip you can move and restyle.

There's also a Create from audio mode that builds a video from a song or voiceover plus a background color or image, captions included. Handy for lyric videos; it deserves its own post, so I'll leave it at that. (If you make music, the vocal extractor pairs well with it.)

The cost, added up

1. Sign up free: 100 tokens and 1 tool use per month, no card.
2. Styling, transcript editing, tap-to-sync, and the customizer: free, no limits.
3. AI auto-captions: 5 tokens per started minute of video.
4. Export: 1 tool use.

So a free account can auto-caption and export one short video a month at zero cost. If you post daily, the Saver plan at $1.99 or a token pack (from $0.99, and purchased tokens never expire) covers a lot of captioning. One honest limit: below the Studio plan, AI transcription caps at 3 minutes per run. Short-form is the target here.

My recommendation: pick one loud style and one clean one, save yourself the scrolling, and let the AI do the timing. Word-level sync is the thing viewers notice, even if they can't name it.