AI-powered song cleaning tool that removes or replaces explicit words while preserving the original singer's voice. Separates vocals from instrumentals, transcribes, detects profanity, and produces clean versions via mute, bleep, or voice-matched replacement.

• Python (pyproject.toml)
• Demucs v4 (Meta, source separation)
• WhisperX (transcription with word-level timestamps)
• Edge-TTS / Bark (voice synthesis)
• Kits.ai (optional cloud voice conversion)
• Gradio (web UI)
• Docker / Docker Compose
• pytest (tests/)

# Install with all extras
pip install -e ".[lyrics,web,ai]"

# Run CLI
sanitune process song.mp3 --mode mute --language en

# Run web UI
sanitune web

# Run tests
pytest tests/

# Run with Docker
docker compose up -d

• src/ — main application source
• tests/ — test suite
• docs/ — documentation and images
• Dockerfile — container build definition
• docker-compose.yml — local development stack

• Audio is processed locally by default; never upload to external services without user opt-in
• AI suggestions require BYO API key (Anthropic/OpenAI); never bundle keys
• Docker image: drumsergio/sanitune with semver tags
• Support both English and Spanish profanity detection
• License: GPL-3.0

Generated by LynxPrompt CLI