Subtwine by CyberMaxAnimated captions for your shorts
Drop a Reel, TikTok or Short. Subtwine writes word-by-word captions with Whisper, lets you fix any word, and burns them in with the style you pick. MP4 + SRT out. It all runs in your browser, so your video is never uploaded.
No sign-up · free for videos up to 60 s · Pro from $9 for 30 days · see Pro
Drop a video here or · or
Check the words
Fix any word, then press Burn in captions. Each line keeps its timing; words are spread across it.
Your captioned video
Subtwine Pro
Pro captions videos up to 30 minutes, exports 1080p, unlocks every caption style (Karaoke sweep, Clean box, Neon) and removes the corner mark. No credits, no per-video limits.
Card checkout by Stripe; your key appears the moment payment finishes, no account. 14-day money-back promise: if it doesn't work as described, email us within 14 days and we fix it or refund you in full (terms). For comparison, Submagic starts at $12/month billed yearly with a monthly video allowance (submagic.co/pricing, 1 Oct 2026).
How to add animated captions to a short video
Drop the video, press Make captions, fix any word in the list, pick a style and position, then Burn in captions. You get an MP4 with the captions in the picture and an SRT file for platforms that take separate subtitles.
Is my video uploaded?
No. Speech recognition (Whisper, via transformers.js) and video encoding (ffmpeg.wasm) run in your browser. The page downloads those open engines once; your video stays on your device.
Why not Submagic, VEED or CapCut?
They are full editors with more effects. Submagic starts at $12/month billed yearly with a set number of videos a month (submagic.co/pricing, 1 Oct 2026) and processes videos on its servers. Subtwine does one job, captions, with no per-video credits, from $9 for 30 days, and nothing is uploaded.
How exact is the word timing?
Whisper times each spoken phrase; Subtwine spreads the words across the phrase by length, which looks right for normal speech. Long pauses inside a phrase can make a word appear slightly early; edit that line if it matters.
How long does it take?
On a laptop without a GPU, plan on roughly a third of the video's length to transcribe with the Fast model, plus about the video's length to render at 720p. With WebGPU it is faster.