Hold a hotkey, speak your mind, release. OiPer transcribes your words in 1.5 seconds — entirely on your machine, without a server in sight.
A single hotkey gesture — press to record, release to transcribe. The most natural input workflow imaginable.
Your voice never travels beyond your machine. Whisper processes everything locally, offline, and privately.
Text injects directly into whatever app has focus. Email, document, message — it just arrives.
CUDA and Metal acceleration available. Or run CPU-only. OiPer adapts to your hardware.
Hook up any LLM — local or remote — for text cleanup. Always your choice, never automatic.
Tiny models for speed, large models for accuracy. Download and swap from a single settings pane.
Native code, zero network latency, GPU acceleration. The combination makes cloud APIs look sluggish.
The Whisper model runs entirely on your hardware. No API, no server, no cloud. Just your CPU or GPU processing your voice.
Logs and audio stay on your machine. You decide what to keep, what to archive, and what to delete.
Text refinement via external APIs is entirely optional. Bring your own key. Default is off. Always opt-in.
Free to download. No subscription. No cloud dependency. Your voice, your hardware, your control.