How to use
- Choose the Spoken language, then drop a video or recording into the box above.
- The AI model downloads the first time, then transcription takes a few minutes depending on length. Progress is shown.
- Check the result, fix any mistakes and tap Save. Choose SRT, VTT or TXT.
Good to know
It runs Whisper, the speech recognition model released by OpenAI, inside your browser. Speech is transcribed sentence by sentence with timing, so you get a subtitle file ready to use.
SRT works almost everywhere — YouTube, Premiere, DaVinci Resolve, VLC — and VTT is the format for web video players. To burn subtitles into the picture, add the SRT to Add Subtitles to Video.
FAQ
How long does it take?
The first time, the AI model has to download (about 100 MB for Faster, 270 MB for More accurate). Processing speed depends on your device: on a computer, Faster takes roughly half the media length and More accurate about the full length or more. Phones are slower.
How accurate is it?
Clear speech is transcribed well, but names, jargon and people talking over each other can be wrong. You can edit the result before saving.
Can I add the subtitles to my video?
Upload the SRT file to YouTube as captions or open it in most video players and editors. To burn the text into the picture, drop the video and the SRT file into Add Subtitles to Video.
Are my recordings uploaded?
No. The speech recognition AI (Whisper) is downloaded to your browser and runs on your own device. Even sensitive meetings and lectures never leave it.
Related guides
Updated: