Audio & video to text
Whisper AI turns speech into a timestamped transcript on your own device, in about 100 languages.
How it works
- 01Drop in a recording, voice memo, podcast or video (MP3, M4A, WAV, MP4, MOV, WebM and more).
- 02Pick the spoken language or let it detect it, then press Transcribe. The first time, the AI model downloads once (about 80 MB).
- 03Watch the text appear, click a timestamp to hear that part, then copy it or download TXT, SRT, VTT or JSON.
Questions
›Is there a limit on minutes or file size?
No. There are no minute caps and no file-size cap from us, because the transcription runs on your own device. Long files simply take longer: on a typical laptop the standard model works at about 1–3× real time on the processor, and much faster on a graphics card with WebGPU. Phones are slower, so they start with a smaller model.
›Which AI model does it use, and what is its license?
Whisper by OpenAI, released under the MIT license, in the ONNX versions published by the onnx-community on Hugging Face. The standard model is Whisper base (about 80 MB on the processor, 149 MB for the graphics-card version); phones use Whisper tiny (44 MB). On computers with WebGPU you can switch to Whisper large-v3 turbo (566 MB), which is noticeably more accurate. The model downloads once, then stays in your browser.
›How accurate is it?
On clear speech it gets most words right, but it's AI, so it can mishear names, jargon and quiet or overlapping voices, and it occasionally invents words during long silences or music. Picking the language instead of auto-detect and using the accurate model both help. It doesn't label speakers: it writes down what's said, not who said it.
›Can it translate a recording into English?
Yes. Turn on "Translate into English" and Whisper writes English text whatever language is spoken. This works with the standard model; the accurate large-v3 turbo model was trained to transcribe only, so it writes the original language.
›Are my files uploaded anywhere?
No. Your files are processed on your own device by your browser. They never touch our servers, so there are no upload waits, no size caps from us, and nothing to delete afterwards.