Record from your microphone or upload an audio file and get an editable plain-text transcription of English audio, powered by Whisper.
Your audio never leaves your device. The English Whisper tiny model downloads once and your browser reuses it on later visits.
English audio only — this page loads the English-only Whisper checkpoint (whisper-tiny.en) and always transcribes English, whatever language you speak.
Runs entirely in your browser. Your data never leaves your device.
When you click Transcribe, four things happen — all locally, all in your browser:
We use Whisper Tiny, the smallest Whisper model (~39M parameters), in its English-only checkpoint. Per the upstream model card it produces good results for clear English audio with one speaker and moderate background noise.
English audio only. This page loads whisper-tiny.en, the English-only Whisper checkpoint, so it transcribes English speech and nothing else — Spanish, Portuguese or Indonesian audio will not work here.
Microphone permissions are requested by your browser when you click Start Recording. The audio is captured locally via the MediaRecorder API (browsers record WebM). Nothing is uploaded to any server.
Yes. Audio longer than 30 seconds is split into consecutive 30-second chunks with no overlap, each chunk is transcribed independently, and the results are joined into one continuous text. No timestamps are produced.
You can upload any audio or video file your browser can decode — the Web Audio API decodes the file in the browser before transcription (MP3 and WAV decode in all major browsers). Microphone recordings come back as WebM. There is no separate audio-extraction step.
If you just finished with AI Speech to Text, the natural next steps are AI Text Summarizer, AI Text to Speech, AI Subtitle Generator, or browse every tool in Local AI Tools.