Local Voice Cloning Director v0.1.0

First public release of Local Voice Cloning Director, a privacy-first Codex Skill for authorized local voice cloning with CosyVoice2.

Highlights

• Local generation with no per-call speech API fee
• Explicit voice-rights and consent gate
• English proper-name pronunciation maps
• Natural, steady, warm, serious, expressive, and energetic delivery profiles
• Style-level prosody guidance without copying a source speaker's identity
• Private provenance manifests and technical audio QA
• English and Chinese documentation
• Codex skill-only plugin manifest with implicit invocation enabled
• Automated public-repository privacy checks and GitHub Actions validation

Install

git clone https://github.com/yjin5562-dev/local-voice-cloning-director.git \  ~/.codex/skills/local-voice-cloning-director

Install CosyVoice2 separately from the official upstream project and follow references/setup.md.

Safety

This release contains no voice sample, voice embedding, model weight, private lexicon, credential, or generated recording. Use only a voice you own or are licensed to clone. Do not use it for impersonation, fraud, authentication bypass, harassment, deceptive political persuasion, or evading synthetic-media disclosure rules.

“No per-call API fee” does not mean zero total cost; hardware, storage, electricity, model downloads, setup, and maintenance remain the operator's responsibility.