Audio to text.
From a voice to an editable transcript. Whisper runs right here, on your device.
Loading workspace…
From input to output.
- Choose an audio file and a language. Listen to check the recording.
- Choose to download the multilingual Whisper model and transcribe locally.
- Review every segment, edit text or speaker annotations, then export TXT, SRT, VTT or JSON.
The useful boundaries.
Up to 50 MB / 15 minutes on desktop; 20 MB / 5 minutes on small-screen devices. WAV is the baseline format; other codecs depend on your browser. Accuracy varies. No automatic speaker identification or cloud fallback.
Your work stays with you.
Files are processed in browser memory. They are not sent to this site’s server, the portfolio assistant, or a business demo. Export downloads a file to your device. Reloading or clearing the workspace removes the active result. Downloaded engine assets may remain in browser cache.
Manage local data · Privacy & data flow
Built with open tools.
OpenAI Whisper · Transformers.js · Xenova Base conversion · Xenova Tiny conversion