AUDIO · TRANSCRIBE
Audio Transcription
Whisper AI transcribes audio directly in your browser. The first run downloads the model (~240MB); subsequent runs use the browser cache.
First run downloads the engine (40–240MB)
- 1.Drag in an audio or video file
- 2.Choose the recognition language
- 3.Click ‘Start Transcription’ (the model downloads on first run)
- 4.Copy the transcript or save it as TXT
→ Output: Transcribed text (saveable as TXT)
The selected model is downloaded once on first run (size grows Fast<Balanced<Accurate), then loads instantly from browser cache.
Upload an audio file and Whisper will convert it to text.
About this tool
Transcribes audio to text with the Whisper speech recognition model. The model runs inside your browser, so the file is never uploaded — which makes it usable for meeting or consultation recordings you cannot send outside.
When to use it
- —Taking the first step from a meeting recording toward minutes
- —Converting an interview to text so you can find quotable passages
- —Transcribing recordings that cannot go to an outside service
Good to know
- —The first run downloads the model and takes time; long files take time to process too.
- —Names, companies, and jargon are frequently wrong. Always check proper nouns against the audio.
- —Overlapping speakers or loud background noise cut accuracy sharply; close-miked, one-at-a-time speech works best.