Help Center

Download the DMG or ZIP from the website, then drag Douya Voice into your Applications folder. On first launch, a setup wizard will guide you through permissions and model downloads.

Two permissions are needed:
• Microphone — for voice recording
• Accessibility — for global hotkey listening and injecting recognized text into other apps' input fields
Grant them in System Settings → Privacy & Security. No restart needed.

After setup, hold the hotkey (default: Fn key) while speaking in any text field, then release to finish. The recognized text is automatically typed at the cursor position.

Try the following:
• Switch the recognition engine (SenseVoice is recommended for Chinese)
• Ensure your microphone works properly and reduce background noise
• Add terms to your Personal Dictionary in Settings to improve proper noun recognition
• Enable AI correction (Pro) to automatically fix recognition errors

Please check:
• Accessibility permission is granted (System Settings → Privacy & Security → Accessibility)
• The hotkey doesn't conflict with other apps
• Try changing the hotkey in Settings
• Restart the app and try again

Voice recognition supports Chinese (Mandarin), English, Japanese, Korean, Cantonese, and more, depending on the engine. The interface supports Simplified Chinese, English, Japanese, Korean, French, German, and Spanish.

Go to Settings → Speech Recognition → Recognition Engine. Choose from Apple Speech, Whisper, SenseVoice, or Paraformer. Each engine has different strengths in speed, accuracy, and language support.

Models are hosted on HuggingFace. Some regions may need a stable connection. Try:
• Check your internet connection
• Retry the download
• If it keeps failing, manually download and import via the Lab page using a URL

The free version includes Apple Speech basic recognition only. Pro unlocks unlimited recognition, AI text correction, all model downloads, smart scene detection, and more.

After purchase, you'll receive an activation key in the format VBAI-XXXX-XXXX-XXXX. Go to the "License Key" page in the app and enter it.

Each activation key supports up to 3 Mac devices. To switch devices, unbind an existing one from the "License Key" page.

Go to the "License Key" page in the app, click "Manage" → "Unbind Device", and confirm. This frees up a device slot for your key.

No. All speech recognition and AI correction run entirely on your device. Audio data never leaves your Mac. This is a core design principle of Douya Voice.

All data (recognition history, settings, model files) is stored locally. History uses a local SwiftData database, and model files are kept in the app's sandbox directory.

Douya Voice uses open-source models like whisper.cpp and SenseVoice for speech recognition, and llama.cpp or Apple MLX for text correction. All inference runs on your Mac's Apple Silicon chip — no internet required.