Privacy Policy
Effective date: November 4, 2026
VoiceInput is a local voice input tool for Windows: hold a hotkey, speak, and when you release it the recognized text is pasted at your cursor.
1. Data collection
VoiceInput does not collect any personal information and does not upload your audio or the recognized text. Speech recognition runs entirely on your computer.
The program creates these files next to its own folder, and they stay on your computer:
- Settings (
config.toml): hotkeys, microphone, dictionary and other preferences. - History (
history.db): recognized text, so you can review and re-paste it from the history window. - Log (
voiceinput.log): for troubleshooting.
Audio is kept in memory only long enough to be recognized; it is never saved as an audio file.
2. Network access
VoiceInput goes online in exactly one case: when you click “Download” in the setup wizard, it downloads the speech model from GitHub (the github.com/k2-fsa/sherpa-onnx releases page). The portable build already includes the model, so you normally never need this step. Recognition itself works offline.
3. Clipboard
To deliver text to your cursor, VoiceInput briefly puts the recognized text on the clipboard, simulates Ctrl+V, and then restores whatever was on the clipboard before. Clipboard contents are never stored or sent anywhere.
4. Third-party services
VoiceInput does not include any analytics, tracking, or cloud storage services.
5. Uninstalling
Delete the program folder and all data is gone. If you enabled “Start with Windows”, also delete the VoiceInput shortcut in your Startup folder (the exact path is listed on the program’s About page).