Your data stays on your device.

Last updated: May 20, 2026.

TL;DR

hush·hush runs entirely on your device. We do not collect, store, or transmit your audio, your transcripts, or anything else. We have no servers, no accounts, no analytics, no advertising, and no third parties involved.

Information we collect

None.

hush·hush has no user accounts, no telemetry, no crash reporting, no analytics SDK, and no network calls of any kind during normal operation. We have no way to know that you use the app, when you use it, or what you dictate.

On-device processing

All speech-to-text transcription and all text organization happens locally on your device, using models bundled inside the app's installer:

Speech recognition uses Whisper (large-v3-turbo) and Qwen3-ASR-0.6B, depending on the language you've selected.
Text organization (punctuation, spacing, capitalization) uses Qwen3.5-4B running under Apple's MLX framework.

Your audio recordings and transcripts stay on your device. They are not uploaded, queued for later upload, or buffered anywhere off your machine. The app makes no network requests during dictation.

What is stored on your device

hush·hush keeps a small amount of state in your macOS user preferences and Application Support directory, all under your control:

Transcript history — a local file of past dictations so you can review or re-paste them. You can clear all history from the History tab.
Custom vocabulary — words you've added or that the app has auto-learned from your corrections, stored locally so future dictations recognize them. You can remove any entry from the Smart Fix tab.
Preferences — your hotkey, primary language, retention settings, and similar options.

None of this is transmitted off your device. You can delete all app data by removing hush·hush from your device and clearing ~/Library/Application Support/hush·hush and the app's entry under ~/Library/Preferences.

Permissions we request

To do its job, hush·hush asks macOS for two permissions. You grant each one through standard macOS system dialogs, and you can revoke either at any time under System Settings → Privacy & Security.

Microphone — required to capture audio while you hold the hotkey. The microphone is opened only during a dictation and is released the moment you let go of the key.
Accessibility — required so the app can listen for your global hotkey and paste the transcribed text into the application you were typing in. The Accessibility permission is not used to read on-screen content, monitor other apps, or capture keystrokes outside of the dictation hotkey path.

Network use

The app does not make network requests as part of dictation. The only outbound network use is when you explicitly click "Check for updates" in the About tab; that action contacts our updates endpoint to check whether a newer version is available. No personal data is sent — only a generic version-check request.

Third parties

None. We do not share data with third parties because we do not collect data in the first place. No advertising networks, no analytics providers, no crash reporters.

Children's privacy

hush·hush is not directed at children under 13, and because the app collects no personal information of any kind, we do not knowingly collect data from anyone, regardless of age.

Your rights

Since hush·hush does not collect or store any personal information about you on our systems, the access / correction / deletion rights established by laws such as the GDPR or CCPA have nothing to act on. All of your data lives on your device, and you can inspect, modify, or delete it at any time using Finder and the app's own controls.

Changes to this policy

If we change this policy, we'll update the Last updated date above and post the revised version at this URL. Material changes will be communicated through the app's release notes.

Contact

Questions? Email hello@hyperpower.dev.