HOLD.
SPEAK.
DONE.

Global hotkey activation. Local Whisper transcription. Direct text injection. Faster than any API. 100% on-device. No cloud, no subscription, no compromise.

1.5s

Transcription 30s audio

Cloud calls by default

2.18×

Faster than nearest API

WHAT IT
DOES

Six things.
All of them matter.

01HOLD. SPEAK. RELEASE.Global hotkey triggers recording. Release to transcribe. Done.

02NEVER LEAVES YOUR MACHINE.Whisper runs locally. Zero cloud calls by default.

03TEXT IN. INSTANTLY.Injected straight into your active app. No clipboard detour.

04GPU OR CPU. YOUR CALL.Auto-detect or force CUDA, Metal, or CPU-only mode.

05LLM CLEANUP. OPT-IN.Any OpenAI-compatible API for text polishing. Your key. Your choice.

06PICK YOUR MODEL.Tiny to Large. Download and switch from the settings panel.

Performance Benchmark

2.18×
FASTER

30 seconds of English audio. Native code vs Python vs Cloud APIs. The result is not close.

◆ No Python overhead

◆ GPU inference (CUDA/Metal)

◆ Zero network round-trips

OIPER DESKTOP1.5s

LEMONFOX API3.27s

PYTHON FASTER-WHISPER3.55s

OPENAI WHISPER 1 API6.46s

YOUR
DATA.
FULL
STOP.

No exceptions. No asterisks. No "unless you opt out."

DEFAULT

Local Transcription

Whisper model runs on your CPU or GPU. No API calls. No server connection. Your audio is processed locally and stays locally.

DEFAULT

On-Device Logs

Activity logs, audio recordings, transcription history — all stored on your machine. You control it. You can delete it.

OPT-IN ONLY

Cloud LLM Cleanup

Optional text refinement via external APIs is 100% opt-in. Needs your API key. Toggle off at any time. Defaults to off.

DOWNLOAD

STOP PAYING.
START OWNING.

No subscription. No cloud lock-in. One download and your voice becomes the fastest input on your machine.

HOLD.SPEAK.DONE.

WHAT ITDOES

2.18×FASTER

YOURDATA.FULLSTOP.