Security & Privacy
Updated: February 7, 2026
Local Processing
Transcription is executed client-side in the browser runtime.
Transcription is executed client-side in the browser runtime.
Media Stays Local
Audio/video content is not sent as transcription payload to our backend.
Audio/video content is not sent as transcription payload to our backend.
Offline-Capable
After model caching, transcription can run without active internet.
After model caching, transcription can run without active internet.
1. Architecture Boundary
OfflineTranscriber uses browser-side inference workers. Service endpoints are used for static assets and licensing/payment flows, not for receiving user media transcription content.
2. Model Downloads and Caching
The first run fetches model artifacts and stores them in browser cache. Clearing browser site data removes model cache and requires re-download.
3. Network Requests Overview
Typical request categories
- Site static assets (HTML/CSS/JS/icons)
- Model artifact downloads
- Optional licensing/payment API endpoints
These requests may include standard web metadata, but not your raw transcription media payload.
4. Operational Security Notes
Keep your browser updated, use trusted devices, and protect local OS accounts. Local-first architecture reduces cloud transfer exposure but local endpoint security remains your responsibility.
5. Related Policies
See Privacy Policy and Terms of Service.