Speech and model management
Multiple model sizes with easy downloads.
OiPer Desktop
Dictation that feels like a shortcut, not a workflow.
Hold a global hotkey, speak your thought, release to inject text into the active app. Designed for privacy, built for speed.
Quick recording
Press, speak, release, done.
Local transcription
Audio stays on-device.
Optional online cleanup
Bring your own key.
Privacy-first architecture.
Everything runs locally by default. Your audio and logs remain on your machine until you decide otherwise.
Local processing
Transcription stays on-device with no hidden uploads.
Online services (optional)
Use your API key for cloud cleanup when you want it.
Fast, seamless workflow.
Designed around the press-and-release hotkey loop.
01
Hold to record
Global hotkey, anywhere on your desktop.
02
Release to transcribe
Text appears instantly in the focused app.
03
Polish if needed
Local cleanup or online LLM refinement.
Configurable without friction.
Pick models, tune backends, and manage providers in minutes.
Backend preferences
Auto detection, CPU mode, GPU acceleration.
Online provider configuration
Custom base URLs, API keys, and model selection.
Advanced options
LLM transcription for technical vocabulary.
Benchmark results.
30 second English sample, end-to-end latency.
OiPer Desktop1.5s
Lemonfox API3.27s
Python Faster-Whisper3.55s
OpenAI Whisper 1 API6.46s
Own the fastest transcription workflow.
Private by default, engineered for performance.