Transcribe audio to text online, free, with no upload
A free online way to turn an MP3 into text, with no sign-up and nothing to install. Drop in an audio or video file, and Whisper turns the speech into text right here in this browser tab. It is private: the speech model is downloaded once and runs on your device, so your audio never leaves it.
MP3 to text, WAV to text, audio or video to SRT and VTT subtitles: one page, all on your device.
Up to 10 minutes per file (5 on a phone). For longer files, many files at once, speaker labels and much faster results, use the free desktop app.
1. Choose a file
2. Choose the speech model and language
Large needs WebGPU, which this browser does not offer. Current Chrome and Edge on a computer have it.
The model is downloaded once from Hugging Face and then kept in this browser.
You seem to be on mobile data or in data-saver mode. The model download is large; Wi-Fi is better.
3. Transcribe
4. Your transcript
Longer files and better accuracy
The free desktop app runs the large Whisper models on your computer: files of any length, a queue, speaker labels, Word and PDF export, YouTube links and live captions. No account and no upload there either.
Download the free appQuestions
Is my audio uploaded?
No. The file is read by this page and transcribed inside your browser. The only download is the speech model itself, from Hugging Face, the first time you use it.
Why is it slower than the desktop app?
A browser tab has limited access to your processor and graphics card. The desktop app runs the full Whisper models at full speed, on files of any length.
Which files work?
Most audio files (MP3, WAV, M4A, OGG) and MP4 or WebM videos with sound, up to 10 minutes. What a browser can open varies; the desktop app opens nearly any format.
Is the model kept, and can I delete it?
It stays in this browser, so the next file needs no download. Remove downloaded models, under step 2, deletes it; clearing this site's data in the browser does too.
Which browsers?
Current Chrome, Edge, Firefox and Safari run Small. Large needs WebGPU: current Chrome or Edge on a computer. On a phone, use Small and a short file.