AI tool · Runs on your device
AI Speech to Text
Transcribe voice notes, recordings and audio files on your device with OpenAI's Whisper model.
or drop a file here · MP3, M4A, WAV, OGG, WebM · up to 10 min
Runs in your browser · audio is not uploaded
How to use it
- Upload an audio file or record from your microphone.
- Choose Fast (about 41 MB download) or Accurate (about 77 MB). The model is cached after the first run.
- Select Transcribe. Edit the transcript, then copy it or download it as plain or timestamped text.
Good for
- Turn a voice memo into notes
- Transcribe a short interview or lecture clip
- Get the text of a voicemail
Privacy and limits
Audio is decoded and transcribed in your browser. Recordings are never uploaded to TodayAI and are discarded when you clear or leave the page.
- Recordings up to 10 minutes and 25 MB
- Not real-time: transcription starts after recording stops and can take a while on slower devices
- Best in English; other languages are detected automatically but accuracy varies
- Timestamps are approximate segment times
Engine: OpenAI Whisper (tiny or base, MIT-licensed; Apache-2.0 ONNX conversion) via Transformers.js. Results can contain mistakes — check anything important.