How to use Subtitle generator
- 1
Open your video
Drop an MP4, MOV, WebM or MKV file (audio files work too). It plays in the preview right away.
- 2
Generate captions
Choose Tiny, Base or Small, keep auto-detect or name the spoken language, and press Generate subtitles.
- 3
Fix and style
Correct words, timings and line breaks next to the live preview; pick a caption style if you want them burned in.
- 4
Export
Download SRT or VTT for upload to YouTube or your editor, or export a new video with the captions in the picture.
Limits and Pro
Files up to 500 MB; the first 10 minutes of each file get subtitles, and burn-in covers the whole video. The Tiny, Base and Small speech models are one-time downloads of 43, 80 and 252 MB.
Files up to 4 GB with up to 600 minutes subtitled per file. See Pro
About Subtitle generator
Subtitle generator turns the speech in a video or audio file into timed captions. A Whisper model (Tiny, Base or Small) runs on your own processor or graphics card, listens to the soundtrack in 30-second windows and writes every word with its timestamp. The words are then grouped into readable caption lines: breaks fall at pauses and punctuation, each line stays under your character limit (42 by default, the Netflix guideline), and by default no caption stays on screen longer than seven seconds.
The editor sits next to a preview that draws the captions exactly as they will be exported. Click a line to jump to it, fix misheard names, nudge start and end times by a tenth of a second, split or merge lines, shift every timing at once to fix a sync offset, and undo any change.
Exports are standard SubRip (.srt), WebVTT (.vtt) or plain text, ready for YouTube Studio, Vimeo, Premiere Pro, DaVinci Resolve, CapCut or VLC. You can also burn the captions into the picture with your own size, colour, background box or outline and position; the video is re-encoded (WebM and MKV become WebM, other formats MP4 with H.264) while the original audio is kept. Speaker names, karaoke-style word animation and translation into languages other than English are not supported.
Privacy
Your video is read from your disk by this page and never uploaded. The audio track is decoded in memory, the Whisper model runs in a background worker on your device, and the burned-in video is assembled locally. The only downloads are the model files from Hugging Face and the media engine from jsDelivr, both cached after the first use.
Features
Private, on-device AI
Speech recognition runs locally with Whisper via WebAssembly or WebGPU. Your video never leaves your computer.
99 languages + translation
Detects the spoken language automatically and can translate speech from any supported language into English subtitles.
Real subtitle editor
Edit text and timings, split and merge lines, shift all timings, follow playback, with undo and redo.
Readable line breaks
Lines are split at natural pauses and punctuation, with your choice of characters per line and one or two lines.
Burn-in captions
Hard-code subtitles into the video with your font size, color, background box or outline and position.
Import existing SRT
Already have subtitles? Import an SRT or VTT file to fix it or burn it into your video.