video1.0<iframe src="https://www.loom.com/embed/66f97cd5d1454346a2b7648eccfa14cc" frameborder="0" width="1920" height="1440" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14401920Loomhttps://www.loom.com14401920https://cdn.loom.com/sessions/thumbnails/66f97cd5d1454346a2b7648eccfa14cc-aa36cae65d22c211.gif420.758In this Loom I built ConsentVault, a governance layer for voice cloning that sits between the LLM and 11Labs text to speech. Voice synthesis is a 2 second API call, and ConsentVault checks consent in about 2 seconds, using cryptographic signed consent scopes. I showed how requests can be blocked, for example when Alex does not consent to political discourse, and how user profiles can stop generation entirely. I also integrated JellyJelly as an identity source, pulling videos and using FFmpeg audio extraction to train voices via 11Labs and PVC. No specific action was requested from viewers.