Tools
Real-time Speech to Text with Whisper (WebGPU)
Mic-to-text captions in your browser using OpenAI Whisper. Prefers WebGPU; falls back to WASM if unavailable.
Tip: Use Chromium 121+ with WebGPU enabled for the best latency.
Model size
Live captions
Idle
Status
Choose a model and click “Download model”.
device: -
model: -
latency: -
How it works
- Runs Whisper in-browser with WebGPU acceleration.
- Captures a few seconds of mic audio, then transcribes.
- Adjust model size for speed vs. quality.
- No audio leaves your device.
If WebGPU is unavailable, the page automatically falls back to WASM (slower).
Transcript
Latest chunks appear at the top.
chunks: 0
Waiting for audio...