# 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.