Audio to text
Free, no sign-up. your file never leaves your device
How to use
Frequently asked questions
It uses Whisper, a transcription AI that runs inside your browser. The first time, the model (40–80 MB, depending on the mode) is downloaded and cached — after that the tool works even offline. Your audio never leaves the device.
It depends on your computer: in "Balanced" mode, roughly the length of the audio; in "Fast", considerably less, with slightly lower accuracy. The bar shows progress chunk by chunk — keep the tab open.
Audio (MP3, M4A, WhatsApp OGG, WAV) and video (MP4, MOV, WebM…) — including recordings from the "Meeting recorder" tool. The text comes out as running prose with punctuation, and you can also download SRT subtitles with timecodes.
No — the result is the text of what was said, without separating voices. Names, acronyms and noisy stretches can come out wrong: review before using.
It does, but transcription is heavy: the phone heats up and takes longer. For long meetings and classes we recommend a computer; on a phone, use "Fast" mode with short audio.