PROJECT_STATE.md — voice-protection

Current Status

Initial project shell created. No application code exists yet.

Objective

Create a Voice Protection Gateway that transforms voice audio into protected audio using watermarking, perturbations, scoring and red-team evaluation.

Durable Rules

Defensive use only.
No voice cloning.
No impersonation.
Protect sensitive audio.
Robustness tests required for transforms.

Known Risks

Misuse for cloning or evasion.
Privacy risk from raw audio retention.
False confidence in protection scoring.

Next Priorities

V-001 Create voice-protection skeleton.
Define audio input/output contract.
Define red-team scoring protocol.

Recent Task Reports

None yet.