Auto-Generate Subtitles and Transcripts From Video — Without Uploading It
Drop a video or an audio file and Toolula listens to it on your device with Whisper, OpenAI's open speech recognition model. The subtitles appear line by line as it works; correct anything it misheard, then download an SRT or WebVTT file, a plain-text transcript, or burn the subtitles straight into the picture. The file never leaves your computer — the speech model (about 50 MB) is downloaded once from toolula.com and then runs in your browser.
How to use it
- Drop a video or audio file.
- Pick the spoken language or leave it on automatic detection, then click Create subtitles.
- Read through the lines and correct what was misheard.
- Download SRT, VTT or TXT, or burn the subtitles into the video.
Frequently asked questions
How accurate are the subtitles?
Toolula uses Whisper base, a compact version of OpenAI's model, so it fits in a browser tab. Clear speech in common languages comes out well; music, crosstalk, heavy accents and rare words cause mistakes. Every line can be corrected before you download it.
How long does it take?
On a recent desktop computer a minute of speech takes about 15 seconds, so a 10-minute video is done in two to three minutes; browsers that allow only one processor core (Safari) take about a third longer. The page shows an estimate for your device before you start. Phones are slower and transcribe at most 15 minutes.
Which languages are supported?
The 99 languages Whisper was trained on, from English, Spanish, German and Chinese to Swahili and Welsh. Detection is automatic, but picking the language yourself is more reliable for short or mixed clips.
Is my video uploaded for transcription?
No. The sound is decoded and recognised inside your browser tab. Only the speech model is downloaded, once, from toolula.com; nothing about your file is sent anywhere.