Speech recognition · runs in your browser

Whisper Tiny Speech Lab

Upload a short English audio clip and transcribe it with OpenAI Whisper Tiny. The model downloads once to your browser cache; audio is processed on your device.

Choose an audio file to begin. First use downloads the model files.

Your transcript will appear here.

About this demo

Uses the quantized ONNX export Xenova/whisper-tiny.en, an ONNX version of openai/whisper-tiny for Transformers.js. English transcription only. Results may contain errors.

Model weights are fetched from the Hugging Face Hub and cached by your browser. Audio is not sent to a server by this app.