Desktop dictation built for speed, focus, and full user control

OiPer turns your voice into text in moments and injects it into the active app. It keeps speech processing local by default, then lets you selectively enable online optimization with your own API key.

Native performanceLocal-first privacy

30-second audio benchmark

OiPer Desktop1.5s

Fastest result with high-quality output and local processing.

Lemonfox API 3.27s · Python Faster-Whisper 3.55s · OpenAI Whisper 1 API 6.46s

User Experience

Hold a global hotkey to talk, release to transcribe, and keep writing without breaking focus.

Performance

Native code, GPU acceleration where available, and very low latency for daily heavy usage.

Privacy

Speech processing runs on-device by default with optional online services only when you decide.

Quick recording flow

1. Hold global hotkey and start speaking.
2. Release key to transcribe immediately.
3. Inject text into your active application.
4. Optionally run local or online text cleanup.

Privacy and security

Transcription, logs, and audio stay on-device by default. No data is sent anywhere unless you explicitly switch on an online provider.

Online optimization is optional and reversible at any time. You control API key usage, base URL, and model settings.

Settings and configuration

Speech model selection with one-click downloads
Backend preferences: auto, cpu, or gpu
Custom base URL, model name, and API key support
Text cleanup using local or online processing
Advanced LLM transcription for technical vocabulary

Accuracy upgrades

Route transcription through an LLM when you need extra precision for technical terms, specialized jargon, or domain-heavy writing. OiPer supports lightweight options like Gemini 2.5 Flash Lite and other compact models.

Local for privacy, online for additional cleanup, or hybrid for the best of both worlds.