A subtitle generator that's actually free
No account, no trial, no forced watermark. The AI runs in your browser, so it costs us nothing to keep it free.
Drop your video here
or click to choose a file · MP4, MOV, WebM · up to 3 min
Choose videoGetting your video ready…
Everything happens on your device. Nothing is uploaded.
Click a caption to edit it. Clear it to delete the line.
Exporting… keep this tab open
Most subtitle tools send your video to their servers, and those servers cost money, so they add watermarks, time limits or subscriptions. Wordflare runs the speech recognition model inside your browser instead. Your device does the work, which is also why your video stays private.
By default a small wordflare.app tag appears in the corner of exported videos. It's how we stay free and it never covers your captions, but it's optional: untick it in the Style tab and your video exports with no branding at all.
Automatic word-by-word subtitles, five animated styles, color, size and position controls, one-click editing of any word, an .srt subtitle file and video export. Free videos can be up to 3 minutes long.
MP4, MOV or WebM, up to 3 minutes. It stays on your device.
Speech is transcribed word by word, in almost any language. Fix any word in one click.
Choose a viral style, adjust size and position, and download your video.
Join the waitlist and get 50% off when it launches.
No. There's no account and no payment details, so nothing can be charged. A future Pro plan will add extra features; the free tool stays free.
Your browser downloads the AI model the first time (about 80–150 MB). After that it's cached and much faster.
About 99 languages, including English, Spanish, Portuguese, French, German, Hindi, Arabic and Japanese.
Yes. Generating, editing and exporting captions is free, with no account and no forced watermark. A small optional wordflare.app tag helps us stay free — you can turn it off.
No. The AI runs inside your browser, so your video and audio never leave your device. The first time, your browser downloads the AI model (about 80–150 MB); after that it is cached.
