AI Video Subtitler

No. Everything runs in your browser. Audio decoding, Whisper AI transcription, caption burn-in — all happen on your device. Nothing leaves your machine.

Whisper Tiny (~40MB) is fast and accurate enough for clear audio. Whisper Base (~150MB) is slower but better for accents, noisy audio, or multiple speakers. Both cache after the first load — you only download them once.

Whisper supports 99 languages, including English, Spanish, French, German, Italian, Portuguese, Russian, Japanese, Korean, Chinese, Arabic, Hindi, and more. Pick the primary spoken language for best accuracy.

SRT is the universal subtitle format — use it for Premiere, Final Cut, DaVinci Resolve, YouTube, and most video editors. VTT (WebVTT) is for web players and HTML5 video. Both are plain text and easy to re-edit.

Burn-in hardcodes the captions directly into the video pixels — great for TikTok, Instagram Reels, and YouTube Shorts, where platforms often ignore uploaded SRT files. Pick a position (top/middle/bottom) and a style (classic box, TikTok outline, minimal shadow), and we render a new MP4 or WebM with the text baked in.

Whisper estimates timestamps from audio chunks and can be ±1-2 seconds off on long clips. Click any timestamp to preview, then click the text to edit. For editors, you can also adjust timings in the downloaded SRT directly.

No hard limit — but longer videos take longer to transcribe (Whisper Tiny runs at roughly 2-5× real-time on modern laptops). For files over 30 minutes, use Whisper Tiny for speed. Memory scales with audio length since the full waveform is held in RAM.

Once the Whisper model is cached (after first use), yes — transcription and burn-in run fully offline. You'd need internet only to reload the page.