Quick press-and-release recording
The interaction is compact enough to stay in the flow of coding, writing, support work, or chat.
OiPer workstation
native / local / configurable
recorder deck
Speak into the workflow, not around it
30 sec benchmark
1.5s
Fast enough to feel immediate while still running locally by default.
1. Hold global hotkey
2. Release to transcribe
3. Inject text into active app
Native execution path
OiPer is designed for low overhead and fast turnaround, with GPU acceleration where the machine allows it.
On-device by default
Speech data, audio, and logs remain local unless the user deliberately enables an online provider.
privacy chassis
Local processing
Audio capture, transcription, and activity history remain on your machine in the default configuration.
Optional online services
Remote cleanup is completely optional and requires your own provider credentials. Turn it on only when it serves the task.
configuration bays
Model rack
Download and switch among speech models sized for speed, quality, or hardware limits.
Backend selector
Run in auto mode, lock to CPU-only, or push to GPU acceleration when available.
Provider dock
Enter base URL, API key, and model name only if you want online cleanup or remote processing.
Precision mode
Use LLM-backed transcription for technical terminology, proper nouns, and industry-specific wording.