How it works
It uses Whisper, an open-source AI model that recognizes speech and turns it into text. It runs inside your browser.
Privacy
Your audio is never uploaded to a server. The first time it downloads the model (about 40 MB), so it takes a little longer.
Tips
- Clear audio without noise gives better results.
- It is a small model: it can make mistakes with strong accents or long recordings.
Frequently asked questions
Is it free?
Yes, no sign-up and no limits.
Is my audio uploaded to the internet?
No, transcription happens on your device.
How accurate is it?
Good with clear speech, but review the text before using it.
Does it work with long audio?
Best with short clips; long ones can take a long time.
Can I transcribe a WhatsApp voice note?
Yes, upload the audio file and choose the language.