Whisper WebGPU

Real-time in-browser speech recognition

How to use this demo:

  1. Click the "Load Model" button to download the Whisper model (~200MB)
  2. Wait for the model to load and initialize (this may take a minute)
  3. Once ready, click "Start Recording" to begin capturing audio
  4. Speak clearly into your microphone
  5. Watch as your speech is transcribed in real-time below

Status: idle

WebGPU Available: No

Debug Info:

Status: idle

Recording: no

Processing: no

WebGPU Available: no

Language: en

Auto Start: no

Ready: no

Loading: no

Stream Available: no

TPS: n/a

Click the button below to load the Whisper speech recognition model. This will download the model (approximately 200MB) and prepare it for use.

Final Transcription:

Speak to see final transcription...