Privacy Policy
Effective November 4, 2026
Nalmeok ("the App") is a desktop application that runs on your computer without user accounts. This policy describes what data the App handles and how it is stored and transmitted.
1. Data stored on your computer
The following data is kept on your computer so the App can work:
- Gemini API keys and app preferences:
config.json - Job history (file names and paths, job status, audio length, API usage and estimated cost):
items.json - Transcripts: the
.mdfile created next to the original, plus a copy the App keeps in thetranscriptsfolder - A troubleshooting log:
nalmeok.log
These are stored in %APPDATA%\nalmeok on Windows (package-private storage for the Microsoft Store version) and ~/.config/nalmeok on macOS. These files are not sent to the developer or any third party. Uninstalling the App or deleting that folder removes it. The .md files created next to your originals are your files and remain after the App is uninstalled.
2. Data sent to Google
To transcribe, the App connects directly to the Google Gemini API using the API key you entered. Each request sends the audio you selected, any reference PDFs in the same folder (used to recognize technical terms), and the transcription instructions.
Google's terms and privacy policy govern how that data is stored and processed. Under Google's terms, data sent on the free tier may be used to improve and develop Google's products, services, and machine learning technologies.
- Gemini API Additional Terms of Service: https://ai.google.dev/gemini-api/terms
- Google Privacy Policy: https://policies.google.com/privacy
3. Information sent to the developer
To tell users whether the Gemini free tier is working well right now, the App sends the outcome of each transcription request anonymously to the developer's server (seook.dev/nalmeok). The results power the free key status page and the App's in-app notices, and are also used as statistics to improve the App (for example, usage by version and operating system).
- What is sent: the request time, free or Batch, the result category (success, quota exceeded, server error, and so on) and HTTP status code, the retry count, whether the App's own usage count showed quota left when a quota error came back, the App version, operating system and CPU type, and a random identifier that is replaced every day
- What is never sent: API keys, file names and paths, audio and PDFs, transcripts, raw error messages
The random identifier is used only to count results from the same app on the same day as one user. It is replaced daily, so the same user cannot be followed across days. When the App checks the status it also sends this identifier so that only other users' failures are counted; the identifier received with that check is not stored. The server uses your IP address only for rate limiting and does not store it. Records are deleted after 30 days.
This is on by default and can be turned off in the App's settings under Send free key status signals. The App contains no advertising identifiers or remote crash reporting.
4. Website analytics
This website (seook.dev/nalmeok) uses Umami, run by the developer at analytics.seook.dev, for visit statistics. It uses no cookies and records pages viewed, referrers, browser, operating system and device type, country, and clicks such as download buttons as aggregate data that does not identify you. These statistics are not linked to the App.
5. Contact
GitHub Issues: https://github.com/jseook11/nalmeok/issues