According to the official disclosure and technical analysis, its data processing mechanism contains:
- Local processing is preferredBasic Dictation Mode: Conversion is done locally in the browser via the Web Speech API, and the audio is not uploaded to the server.
- Cloud Encryption: AI mode requires uploading contextual data and voice clips, using TLS 1.3 transmission and AES-256 storage encryption
- Data retention policy: Original audio retained for up to 72 hours for quality optimization, transcribed text automatically anonymized after 30 days
User controllable permissions include:
- History can be cleared in the settings panel
- Disable the "Improve recognition accuracy" option to stop data collection
- Use stealth mode to avoid storing session logs
Safety Recommendations:
Avoid entering sensitive information such as credit card numbers in the plugin
Enterprise users can apply for deployment of a private speech recognition engine
Periodically review Chrome's microphone permission settings (chrome://settings/content/microphone)
The plug-in is ISO 27001 certified for information security management, but a compliance assessment is still recommended before use in sensitive industries such as medical and legal.
This answer comes from the articleVoquill: Browser Plugin for Converting Speech to TextThe
