Legal

Privacy Policy

Last updated October 28, 2026

The App

VoiceCatcher is a privacy-first voice transcription application. All audio processing happens locally on your device. VoiceCatcher never sends your audio or text anywhere. (If you use Apple’s Universal Clipboard, macOS itself can offer what VoiceCatcher puts on your clipboard to your other Apple devices; turn Handoff off to keep it on this Mac.)

Data Processing
  • Audio recording: Captured from your microphone and processed entirely on-device. Audio is never transmitted to any external server.
  • Instant-capture buffer: So the first word of a recording isn’t clipped while the app starts capturing, VoiceCatcher keeps a short rolling buffer (about 1.5 seconds) of the most recent microphone audio in memory only. It is continuously overwritten, is used solely to prepend that lead-in to a recording you start, and is never written to disk or transmitted.
  • Transcription: Performed locally using the Whisper AI model on your CPU or GPU. No cloud-based transcription services are used.
  • Post-processing: Text formatting (punctuation, capitalization, paragraph breaks) is performed locally in the app.
  • The ⌘⌘ gesture listener: To detect the double-tap recording shortcut, VoiceCatcher registers a listen-only system event listener for modifier-key transitions and a bare “some key was pressed” signal. It never reads which key you typed, never reads text, and never touches your clipboard — it only compares modifier keycodes and timestamps. It uses the same macOS Accessibility permission auto-paste already requires; no additional permission is requested.
  • A meeting’s name, from the call’s window: During a meeting, VoiceCatcher reads the titles of the call app’s windows and, in a browser, its tab titles (never the pages), about every 30 seconds, with the Accessibility permission ⌘⌘ already uses. Only a title that names the meeting is kept, with that meeting in your History. Nothing else read is stored, logged, or sent anywhere.
  • Screenshots and screen recordings (off until you switch them on): VoiceCatcher only sees your screen when you take a screenshot or record a GIF or video, with the screenshot gesture or the menu-bar item. A recording sees only the area or window you choose, until you stop it. Nothing is captured in the background or on a timer. Each picture and recording is saved on your Mac and never uploaded.
  • Your calendar (off until you switch it on): With Meetings: names from my calendar on, VoiceCatcher reads the Mac’s Calendar when a meeting starts. Only the event happening then is kept, as its title and the attendees’ names, with that meeting; email addresses aren’t kept. It only reads; nothing is changed or sent. macOS asks once, when you switch it on.
Data Storage
  • Session history: Transcriptions are stored in a local SQLite database on your Mac. You can delete individual sessions in the app at any time; to delete everything, see Data Deletion below.
  • Settings: Stored as a local JSON file in the app’s data directory.
  • Recordings’ audio: Once a recording has been transcribed, its audio is kept for 30 days by default (Settings → Window & storage: 7, 30 or 90 days, or forever) and then deleted from your Mac; the transcript and its screenshots are kept.
  • Audio checkpoints: Temporary audio files used for crash recovery, stored locally and cleaned up after use.
  • Log file: Technical events go to ~/Library/Logs/VoiceCatcher, the last 7 days kept. No transcript text and no screenshot names.
Network Requests
VoiceCatcher never sends your audio, transcripts, screenshots or settings anywhere. Its only requests: updates from voicecatcher.io, checked at launch and once a day, installed only if they carry VoiceCatcher’s signature and never during a recording (nothing about you is sent; you can turn it off in Settings → Window & storage); the Large speech model, downloaded from Hugging Face only if you ask; and, if the built-in model file is ever missing or damaged, the same model once from Hugging Face, only if you choose Repair when VoiceCatcher asks (it never downloads it by itself). Downloads are checked against known checksums. A report (Diagnostics → Send a report) is only ever sent by you, from your own email.
What We Do NOT Collect
No telemetry, analytics, crash reports, usage tracking, advertising identifiers, personal information, or transmitted audio/transcription data.
Data Deletion
All data is stored locally. Delete individual sessions in Session History, or remove everything by deleting the app’s data directory (~/Library/Application Support/com.harte.voicecatcher/). Uninstalling removes the app but not the data directory.
Contact
VoiceCatcher is made by Ngentix, Inc. Privacy questions: voicecatcher@ngentix.ai.

This Website

Analytics

voicecatcher.io uses Google Analytics to count visits and waitlist sign-ups. It sets cookies in your browser. Google receives the pages you view, the page that sent you here, the links you click to leave the site, your browser and device type, and your IP address, which it uses to estimate your approximate location. When you join the waitlist, Google is told that someone signed up, never your email address. We use it only to see how many people visit and how many join the waitlist. To opt out, install Google’s opt-out browser add-on or use a browser or extension that blocks trackers.

The waitlist

There is nothing to buy yet. To hear when VoiceCatcher goes on sale, you give us your email address through a form on this site. The form is handled by Netlify, which hosts voicecatcher.io and keeps a copy of what you send; the address then reaches Ngentix, Inc. by email. We use it to tell you when VoiceCatcher is on sale and to reply to you. To have your address deleted, email us at the address below.

Contact

Privacy questions: voicecatcher@ngentix.ai