Privacy-First Voice-To-Text

SPEAK.

TRANSCRIBE.

LOCALLY.

Hold a hotkey. Speak. Release. Your words appear instantly in any application. No cloud. No data collection. No compromise.

1.5

Seconds / 30s audio

100% Local

Zero cloud dependency

GPU Accelerated

Native performance

01 / Features

001

HOLD TO RECORD

Press your global hotkey to start. Release to transcribe. Text is injected directly into your active application with zero friction.

002

LOCAL PROCESSING

Transcription runs entirely on your machine. Audio and logs never leave your device. Privacy is the default, not an option.

003

NATIVE SPEED

Written in native code with GPU acceleration. Transcribes 30 seconds of speech in 1.5 seconds. No Python. No Electron.

004

FULL CONTROL

Choose your backend (auto/CPU/GPU). Configure API keys. Select models. Local or online text cleanup. Your workflow, your rules.

02 / Performance

BENCHMARK: 30S ENGLISH AUDIO

Raw transcription speed comparison. OiPer runs natively with GPU acceleration — no Python runtime, no API latency.

OIPER DESKTOP1.5s

LEMONFOX API3.27s

PYTHON FASTER-WHISPER3.55s

OPENAI WHISPER 1 API6.46s

03 / Privacy

LOCAL BY DEFAULT

Transcription runs entirely on your machine
Activity logs and audio stay on-device
No data sent anywhere without your choice

ONLINE: OPTIONAL

Your API keys. Your decision.

Online optimization is completely opt-in
Uses your own API key — we have no access
Toggle on or off at any time

04 / Configuration

CFG.01

SPEECH MODELS

Download and select models sized for your hardware.

CFG.02

BACKEND

Auto / CPU / GPU — choose your compute path.

CFG.03

LLM ROUTING

Gemini Flash, custom models for extreme accuracy.

CFG.04

API CONFIG

Custom URLs, keys, model names. Full control.

CFG.05

TEXT CLEANUP

Local or online post-processing of transcriptions.

CFG.06

TERMINOLOGY

Fine-tune for technical and specialized content.