Turn a recording into a transcript you can use.
Make an interview searchable, turn a voice memo into a draft or find a quote in a recorded lecture. Whisper AI runs speech recognition in your browser and gives you editable text with timed cues. The free local workflow needs no account.
From audio to text in three steps
- Choose Upload a file above, or use Record audio for a new recording. Start with a file under 200 MB and 20 minutes.
- Select the spoken language and choose Start transcription. The first run downloads the model; keep the page open while it processes.
- Listen back, correct the draft and select TXT in Export format. Choose Export to save your text.
Choose the output for your next step
TXT is a plain reading copy. SRT and VTT retain timestamps for subtitles. JSON keeps structured cue data, while Word provides an editable document. You can change cue text, adjust timing and add a missed line before exporting.
A private workflow with practical limits
Your media stays on your device in local mode, although model downloads need a network connection. Speed and accuracy depend on the recording and your hardware. If the browser struggles, try Tiny in Advanced settings and a shorter MP3 or WAV excerpt. Local history belongs to this browser; export a copy before clearing its storage.
Working from a video instead? Use video to text. Need captions for a player? Try the subtitle generator.