Tools › Audio
Transcribe audio to text
Turn interviews, lectures, podcasts and voice memos into text with timestamps. Whisper runs in your browser, so the recording never leaves your device. Download TXT, SRT or VTT.
The speech model downloads once to your device; your recording never leaves it. Nothing starts until you choose a file.
How it works
- Drop an MP3, M4A, WAV, OGG or a video. The sound is decoded on this device and resampled to 16 kHz mono for Whisper.
- Long recordings are cut into parts of up to 30 seconds at pauses, so a one-hour interview is fine. Text appears part by part with an estimate of the time left.
- Correct any word or timestamp in place, then copy the text or download it as TXT, with or without times.
Questions
Which languages work?
All 99 languages Whisper knows, from English, Spanish and Hindi to Japanese, Arabic and Turkish. The language is detected automatically, or you can choose it.
Is it really private?
Yes. The model is downloaded once and runs in this tab. Nothing about your recording is sent anywhere.
Can it tell who is speaking?
Not yet. You get the words with timestamps; speaker labels are not added.
How accurate is it?
Clear speech comes out close to word for word. Background music, crosstalk and heavy accents cause more mistakes; the Accurate model helps.