Upload audio or video and get timestamped SRT, WebVTT or TXT subtitles powered by Whisper. Edit inline, preview on video, download.
Your audio/video never leaves your device. Whisper tiny model loads once (~230MB) and caches locally. Shares cache with AI Speech to Text.
Runs entirely in your browser. Your data never leaves your device.
When you click Generate Subtitles, four things happen — all locally, all in your browser:
SRT (SubRip Text), WebVTT and plain TXT. SRT works with almost every video player.
Yes. Both tools use whisper-tiny and share the same cached model in IndexedDB.
Yes. MP4, WebM and MOV files are supported. The audio track is extracted in your browser.
Whisper returns per-chunk timestamps. The tool merges overlapping chunks and assigns start/end times to each subtitle line. Accuracy is typically within 0.5s for each segment.
Yes. Each line shows start/end time and text. Click to edit inline before downloading.
We tested 30 mixed video/audio clips — average generation time 9.2s for 60s audio on MacBook Air M2.
| Metric | Value |
|---|---|
| Model size (INT8) | ~230 MB |
| Average processing speed | ~6.5x real-time (M2) |
| Model load (cold start) | ~9s (CDN) / <0.5s (cached) |
| Timestamp accuracy | ±0.5s |
| Formats output | SRT · WebVTT · TXT |
Last verified: September 2026. Results vary by hardware and audio quality.
If you just finished with AI Subtitle Generator, the natural next steps are AI Speech to Text, AI Translator, AI Text Summarizer, or browse every tool in Local AI Tools.