Transcribe audio

Turn speech into text with AI that runs on your device. Your recordings are never uploaded.

Drop an audio or video file

Runs on your device with on-device AI. Your files are never uploaded.

How it works

  1. Drop an audio or video file onto the page, or choose one from your device.
  2. The first time, the AI model downloads and is cached for next time. Then watch the transcript appear.
  3. Copy the text, or download it as .txt, .srt or .vtt.

Questions

Is my recording uploaded anywhere?
No. The AI model runs in your browser, on your device. Your audio, its name and the transcript never leave your device and are never sent to a server.
Why does it download a model the first time?
The AI model is about 80 MB, and the engine that runs it 14 to 27 MB, depending on your device. They are downloaded once and cached in your browser, so later visits start straight away. The download contains none of your data.
How long can the audio be?
Up to 30 minutes, in a file of up to 300 MB, for now. Your device does the work, so it can take about as long as the audio on slower devices. The transcript appears as it goes.
Which files and languages work?
Common audio formats such as MP3, M4A, WAV and OGG, and the sound in MP4, MOV and WebM videos where your browser can read it. The language is detected for you, and accuracy is best for English.
What are .srt and .vtt files?
They are caption files with timestamps. Video editors and players such as YouTube can use them as subtitles. Use .txt if you only want the words.

Related tools