Global Hotkey
Press-and-hold activation system-wide. Works across every app without losing context.
Local Inference
Whisper runs on your GPU. Zero egress. Audio never leaves your machine by default.
GPU Accelerated
CUDA and Metal support. Auto-detect your hardware or manually override the backend.
Direct Injection
Transcribed text appears in your focused input instantly. No clipboard required.
LLM Cleanup
Optional text polish via any LLM — local or cloud. Always your key, always your choice.
Model Manager
Download and switch Whisper models in-app. Tiny for speed, Large for precision.