Transcribe Video to Text and Subtitles
Write the subtitles from the speech in a recording, on your own machine.
An auto subtitle generator that transcribes the speech in a video into SRT, VTT or plain text, on your own machine. No upload, no per-minute charge.
What this tool is for
The site could already burn subtitles in, pull them out, convert and resync them — everything except write them. This is the missing piece, and it is also the one with the strongest reason to run on your own machine. Think about what people actually transcribe: a client meeting, an interview, a deposition, a doctor’s notes, a journalist’s source recording. Every service that does this wants the recording uploaded first, which for most of those files is exactly the thing you cannot do.
How accurate it is
Clear speech with little background noise transcribes very well. Music, crosstalk and strong accents reduce accuracy. Always read through the result before publishing it.
What to do with the file
Upload the SRT alongside a video on YouTube or Vimeo, load it in a player, or burn it into the picture with Burn In Subtitles so it shows everywhere, including on social apps.
Why it is private
The speech model runs on your own device. Interviews, meetings and unreleased footage are never sent to a transcription service.
How to use it
- Add your files. Drop them on the page or choose them. They stay on your device.
- Set the output format and quality. The defaults suit most files.
- Press convert and save the results, individually or as one ZIP.
Questions
How do I get subtitles for a video without uploading it?
Use this. The speech model runs inside the page, the same way the video engine does, so the recording is read on your own processor and never sent anywhere. There is no upload endpoint to send it to.
How accurate is it?
It is Whisper tiny, which is very good on clear speech in a quiet room and gets worse with heavy accents, crosstalk and background noise. Treat the output as a first draft to correct rather than a finished transcript, which is true of every automatic system including the paid ones.
How large is the download?
About 39 MB the first time, then it is cached. That is the same order as the video engine this site already downloads, and it is what running a speech model locally costs. Nothing is downloaded at all unless you use this tool.
How long does it take?
Faster than the recording itself on a normal laptop — roughly four times faster in testing — though a phone or an older machine will be slower. It works through the audio in thirty-second windows with a small overlap so a word cannot fall into a gap.
What formats does it write?
SubRip (.srt), which is what most players and editors expect; WebVTT (.vtt) for subtitles on a web page; or plain text if you only want the words. Burn In Subtitles will then put an .srt permanently onto the picture.
Does it handle other languages?
This model is the English-only build, which is why it is small and quick. Speech in another language will come out as nonsense rather than as a translation.