OiPer is a privacy-first voice-to-text desktop app. Hold a global hotkey to record, release to transcribe, and the result is injected into the active app. Optional text cleanup can run locally or via an online provider.
- Download the latest release from the GitHub Releases page for this repository: https://github.com/oiper/desktop/releases
- Install the package for your OS.
- Availability: Only Windows is supported now. macOS support is coming soon.
- On first launch, allow microphone access when prompted.
- Open OiPer.
- Choose a speech model and download it.
- Press and hold
Shift+Spaceto record. - Release to transcribe and insert into the focused app.
- Speech model selection and downloads.
- Backend preference (auto/cpu/gpu, where available).
- Text optimization: online or local.
- Online provider configuration (base URL, API key, model name).
- Autostart on system login.
- Transcription runs locally on your machine.
- Online optimization is optional and requires your API key.
- Activity logs and audio stay on-device.
During our development phase, before the official release, the app is not yet signed with Apple. If macOS blocks the app or reports it as damaged, run the following command in Terminal to remove the quarantine attribute:
xattr -rd com.apple.quarantine /Applications/OiPer.appWe are working on signing the app with Apple, after which this step will no longer be needed.
After updating the app, you may need to grant permissions again from scratch. Even if the app already appears as allowed in the system permissions list (for example, Microphone or Accessibility), it may not actually work. In that case, remove the existing entry and add it again.
- Use GitHub Issues to report bugs and request features: https://github.com/oiper/desktop/issues
- Use GitHub Discussions to ask questions and share ideas: https://github.com/oiper/desktop/discussions
- Check the Releases page for updates: https://github.com/oiper/desktop/releases