Granite Speech • WebGPU • Transformers.js

Granite Speech WebGPU

Load the model once, then transcribe audio directly in the browser with WebGPU acceleration. When initialization finishes, the page will take you straight to the input workspace.

Input

Audio

No audio selected yet.

Output

Transcript

Load the model, then choose an audio file or record a clip.

The transcript will stream here as it is generated.