stage 01
Capture
A global hotkey arms recording immediately from anywhere in the desktop environment.
OiPer Desktop
privacy-first signal chain
The local transcription engine that keeps pace with thought
OiPer is built for people who want voice input to behave like an extension of the keyboard. Record by holding a hotkey, release to transcribe, then let the text appear directly inside the app in focus.
stage 02
Transcribe
Release the key and OiPer pushes the clip through the chosen local backend with low latency.
stage 03
Inject
Text appears in the focused application so the spoken sentence becomes part of the working draft.
stage 04
Polish
Choose local cleanup or optional online optimization when the task calls for it.
User experience
The product gets out of the way. The interaction is short enough to feel closer to a keyboard shortcut than a recording session.
Performance
Native implementation, backend selection, and GPU acceleration where available keep turnaround fast.
Privacy
Speech stays on the machine unless the user explicitly enables a provider and supplies their own credentials.
benchmark wall
30-second audio
Lowest time wins. OiPer finishes first.
1.5s
OiPer Desktop
3.27s
Lemonfox API
3.55s
Python Faster-Whisper
6.46s
OpenAI Whisper 1 API
privacy and services
Local processing
Audio and logs remain on-device. The default route keeps data entirely on your machine.
Online options
Enable cleanup only if you want it. Supply your own API key, base URL, and model choice.
configuration matrix
Speech model management with easy downloads
Backend choice across auto, CPU, and GPU
Custom base URL, API key, and model fields
Optional online cleanup and local rewrite paths
LLM transcription for high-precision terminology