Privacy Framework
How we secure, process, and respect the unique architecture of your digital voice identity.
Effective Date: January 24, 2026
Core Principles
1. Biometric Data Collection
Voxtar collects voice biometric data ('Voice Prints') solely for the purpose of creating your persistent AI voice model. This data is converted into numerical vector embeddings immediately upon ingestion. We do NOT store raw audio files longer than necessary for the initial training process (typically 24 hours).
2. Vector Embeddings Security
Your Voice Print is stored as an encrypted vector embedding (a mathematical representation) in our secure architectural database. These embeddings cannot be reversibly engineered into original audio files, ensuring that your raw voice identity remains protected even in the event of a data breach.
3. Use of Data
We use your data to: (a) Provide the voice synthesis service; (b) Improve the accuracy of our neural rendering engine (using anonymized, aggregated data only); (c) Prevent fraud and deepfake abuse through our 'Safety-First' watermark ledger.
4. Third-Party Processors
We partner with trusted cloud infrastructure providers (e.g., AWS, Azure) to host our GPU clusters. These providers are SOC 2 compliant and contractually obligated to process data only according to our strict architectural standards. We do not sell your data to advertising networks.
5. User Rights & Deletion
You retain full ownership of your Voice Print. You may request the deletion of your account and all associated embeddings at any time via the Dashboard settings. Upon deletion, your data is cryptographically erased from all active and backup systems within 30 days.
Your voice is your data. Use it securely.