How to create SRT
- 01
Choose one video
The browser validates its signature, duration, and size before starting.
- 02
Generate and edit cues
Select the spoken language, run local transcription, then fix wording, timing, and line length.
- 03
Export SRT
Resolve blocking timing issues, download SRT, and test it against the original video.
Supported workflow
- MP4, MOV, M4V, or WebM video within the published budget
- Up to 200 MB and 60 minutes
- SRT, VTT, TXT, and JSON export after cue review
Subtitle limits
- Speech recognition can mishear words; cue timing and line breaks also need review.
- Some browser and codec combinations cannot decode a valid-looking video.
- Large media can use significant memory and requires the tab to remain open.
Check subtitles in context
Review the first, middle, and final minute with the SRT enabled in a real player.
- Confirm the first cue starts at the right moment.
- Check names and numbers against the audio.
- Look for overlaps and captions that remain too long.
If generation fails
- Confirm the video plays with sound in the same browser.
- Try a shorter video or a common MP4/WebM encoding.
- Close memory-heavy tabs and allow the documented model and runtime requests.
Questions about video to SRT
Does the video leave the browser?
No. Media decoding, speech recognition, cue editing, and export run locally. Runtime and model downloads are separate network requests.
Are generated timings final?
No. Review timing and wording against playback before publishing or translating the subtitles.