Privacy
Your work stays
on your device.
Mergonomy Voice is local-first. Voice, transcripts, notes, meetings, and imported media are not stored on Mergonomy servers.
Last updated: September 12, 2026
Transcription, cleanup, media tools, notes, and meetings run and remain on your device.
Mergonomy servers do not save microphone, meeting, or imported-media audio.
Delete individual notes and meetings, or use Factory reset to remove Mergonomy Voice-owned data.
When enabled, Mergonomy Voice sends only allowlisted feature-use events and day-wise Speak to Type counts—never your content.
1. Scope
This notice covers the Mergonomy Voice Windows app and the Mergonomy Voice pages on Mergonomy.com. It describes how the current local-first release handles data.
2. Windows requirements
- Supported systems: Windows 10 or Windows 11 on an x64 computer. Windows on ARM is not currently supported.
- Microphone: microphone permission is required for dictation, voice notes, and the local side of meeting capture.
- Meeting capture: Windows system-audio capture is used to transcribe the other side of a meeting.
- On-device speech model: the current multilingual Whisper model uses about 57 MB of disk space and approximately 650 MB of memory while running.
- Optional cleanup model: depending on the model selected, allow about 378 MB to 1.03 GB of disk space and approximately 700 MB to 1.8 GB of memory.
3. Data stored on your device
Mergonomy Voice stores its working data in Electron's per-user application data directory.
On Windows this is typically %APPDATA%\m-voice or %APPDATA%\mergonomy-voice.
- Voice audio
- Microphone and meeting audio are used to create a transcript. Voice recordings are not kept as product data or saved on Mergonomy servers.
- Notes and transcripts
- Voice notes, rich-text content, timestamps, language, and related metadata are saved as local files.
- Meetings
- Each meeting's transcript and notes are stored in its own local folder with a backup copy to reduce the effect of file corruption.
- Media transcription history
- Mergonomy Voice stores the source file path, transcript segments, and subtitle edits. The original audio or video remains in the location you selected and is not copied into Mergonomy Voice's app-data folder.
- Models and settings
- Downloaded speech and cleanup models, app preferences, usage counters, custom words, and your selected activation sound are stored locally.
- Diagnostic logs
- Error and diagnostic logs are stored locally, redact known sensitive fields, rotate at 2 MB with up to three archives, and expire after seven days. Diagnostics leave the device only when you explicitly export and share them.
4. How your data is processed
Local features
Speech recognition, transcript cleanup, imported audio and video transcription, subtitle generation, notes, meeting transcription, meeting data, and meeting notes are processed locally. Their content does not leave your device through these features.
Model downloads
When you choose to download an on-device model, Mergonomy Voice connects directly to the model's third-party host, currently Hugging Face or GitHub. The host receives the network information normally required to deliver a file, such as your IP address and request metadata. Mergonomy Voice does not send voice audio, transcripts, notes, meetings, imported media, filenames, custom words, or account details with a model download.
5. Retention and deletion
- Local content: notes, meetings, media history, models, settings, and usage history have no automatic expiry. They remain until you delete them or run Factory reset.
- Delete a note: permanently deletes that note's local file.
- Delete a meeting: permanently removes that meeting's local folder, transcript, and notes.
- Factory reset: removes Mergonomy Voice-owned notes, meetings, histories, downloaded models, settings, logs, and the custom activation sound from the device.
- Logs: expire after seven days as described above.
- Pending analytics: unsent day-wise Speak to Type counts remain on your device until Mergonomy Voice next goes online. Disabling analytics immediately removes those counts and any queued analytics from the device.
- Exports: files you exported elsewhere remain until you delete them yourself.
6. Uninstall behavior
Uninstalling Mergonomy Voice removes the application but intentionally preserves local app data, including notes, meetings, settings, models, and logs. To remove Mergonomy Voice-owned data, use Settings → Account & Model → Factory reset before uninstalling.
Files you exported elsewhere are never removed by Factory reset or uninstall.
7. Meeting consent is your responsibility
You are responsible for deciding whether a meeting may be recorded or transcribed and for giving any notice or obtaining any consent required by law, contract, workplace policy, or platform rules. Do not record or transcribe a meeting unless you have the necessary permission.
8. Optional pseudonymous product analytics
After onboarding is complete, Mergonomy Voice enables limited product analytics to help Mergonomy understand which areas of the app are useful and improve the health of the product. You can disable analytics at any time in Settings → About → Usage analytics. No analytics event is sent merely because the app opens, onboarding is shown, or analytics is enabled or disabled.
When analytics are enabled, Mergonomy Voice sends only a randomly generated pseudonymous app identifier, the event name and time, and the following allowlisted information when it applies:
- the day-wise number of completed Speak to Type uses;
- Home actions to start a meeting or create a note;
- Meeting actions to open the meeting folder or start a meeting;
- Speak to Type actions to open Settings or History or run the Accessibility check;
- Notes actions to view metrics or create a new note; and
- Audio or video media actions to select a tab, copy a transcript, save to notes, create subtitles, use Focus mode, transcribe another file, or download a plain or timestamped transcript.
Speak to Type counts are stored locally by calendar day while the device is offline and are sent when it next goes online. A day with no Speak to Type use produces no daily-count event. Disabling analytics stops new collection immediately and removes unsent daily counts and queued analytics from the device.
Analytics never includes voice audio, transcripts, notes, meeting content, meeting notes, imported media, filenames, file paths, custom words, exported files, diagnostic logs, or crash reports. Automatic page views, session replay, heatmaps, error tracking, advertising profiles, and precise location collection are disabled.
PostHog Cloud processes and stores these events in Mergonomy's access-controlled project on Mergonomy's behalf. Person profiles and IP-based geolocation are disabled. Mergonomy uses analytics only to understand product usage and improve M Voice—not for marketing, advertising, personalization, cross-service tracking, or the sale of data.
Turning analytics off does not automatically delete events already received. For a privacy question or deletion request, use the contact form.
9. Mergonomy Voice website data
The website host may process standard network and server-log information needed to deliver and secure the site. The Mergonomy Voice pages on Mergonomy.com also load the Space Grotesk and Inter fonts and Google Analytics scripts from Google's servers, which causes your browser to request them from Google.
10. Support and privacy questions
Use the contact form for privacy questions. To share product feedback or a support query, use the feedback form.
11. Changes to this notice
This notice may change if M Cloud is introduced. Any updated notice will be published before the changes take effect. The updated date at the top identifies the current version.