Transcribe Audio to Text
Turn a recording or a video into text with Whisper running in your browser — Italian, English and more — with subtitles to save as .srt.
What it does
Transcribe Audio to Text works on MP3, WAV, M4A, OGG, FLAC, AAC, WEBM, MP4 and MOV with ffmpeg and audio engines compiled for the browser, so the recording or video is processed by your own computer or phone — the same programs a desktop editor uses, without installing one.
When you need it
Usually right before sending or publishing: a video too heavy for WhatsApp, a voice note full of hum, a clip that needs captions because the feed plays it muted, an interview to turn into text.
Limits and caveats
Long or high-resolution files take as long as your device needs, and a phone is slower than a laptop. The engines download once, the first time, and come from KIVO itself. The original file is never changed: you always download a new one. Everything happens in your browser, so the file never leaves your device.
How to transcribe audio to text
- Pick the language spoken, or let it detect it.
- Drop a recording or a video.
- Copy the text, or save it as .txt or as subtitles.
Frequently asked questions
- Is the recording uploaded?
- No. The speech model runs in your browser, so a call, a meeting or a voice note never leaves your device. It is downloaded once (about 41 MB) and kept.
- How accurate is it?
- It is Whisper's small model, chosen so it runs on a phone. Clear speech comes out well; names, jargon, accents and noisy rooms less so. Read it through before relying on it.
- How long can the recording be?
- There is no limit as such: it works through it thirty seconds at a time. An hour takes several minutes on a computer and longer on a phone.