Hold a hotkey. Speak. Release. OiPer transcribes with native-code speed and injects text directly into your active window — zero cloud, zero latency, total privacy.
Any key combo, any application. Hold to record, release to transcribe. Zero switching required.
Transcribed text is placed directly into your focused window. No clipboard gymnastics.
Whisper runs entirely on your hardware. Multiple model sizes from ultra-fast to ultra-accurate.
Auto-detects CUDA and hardware acceleration. Falls back gracefully to CPU mode.
Optional LLM-powered text polishing. Use local models or your own cloud API key.
Backend, model size, language, cleanup mode. Every parameter is configurable.
All tests: 30 seconds of English audio · time measured to transcription complete
Every parameter is exposed. Choose your model, backend, language, and optimization mode. OiPer adapts to your hardware and workflow.
All speech-to-text processing runs on-device via local Whisper models. No network calls, no exceptions.
Audio recordings and transcription logs are stored only on your machine. We have zero access.
API-based features require your own key and explicit activation. Disabled by default, always.
No analytics, no crash reporting, no usage monitoring. Your workflow is entirely private.
Download OiPer and experience voice transcription at native-code speed — private by design, powerful by default.