Frequently Asked Questions

No. Airgap Voice runs 100% locally on your Mac. Both models are bundled with the application - no download is ever required. The app doesn’t even have the macOS entitlement to open network sockets - it physically cannot connect to the internet.

Airgap Voice supports over 30 languages including English, Chinese (Mandarin and Cantonese), Spanish, French, German, Japanese, Korean, Hindi, Arabic, Portuguese, Russian, Italian, Dutch, Swedish, Danish, Finnish, and many more. The Precision model supports the widest range of languages. Enable Auto-Detect in Language settings to automatically identify the spoken language.

No. Airgap Voice does not write audio recordings, transcripts, or usage logs to disk. All audio is processed in volatile RAM and zeroed out after each session. Only user preferences (microphone selection, keyboard shortcut, formatting settings) are persisted. There is zero forensic footprint.

Turbo is best for everyday dictation in English and European languages. It’s faster and works well on any Apple Silicon Mac.

Precision delivers the highest accuracy and is recommended for: Asian languages, complex multilingual content, specialized terminology, or when maximum accuracy matters more than speed. It requires an initial download (~2 GB) and uses more GPU memory.

Airgap Voice is designed for HIPAA-compliant environments. Since no patient data ever leaves your device - no cloud processing, no network transmission, no disk storage of transcriptions - the primary vectors for HIPAA violations are eliminated by architecture. The application runs in a macOS App Sandbox without network entitlements, and all audio buffers are zeroed after use. However, overall HIPAA compliance depends on your organization’s complete security posture, not just the transcription tool.