LICENSE_COMPLIANCE.md

📋 Open Source License Compliance Report

Current Project License

• Clementine: GNU GPL v3 (GPLv3)
• File: COPYING
• Requirement: All dependencies must be GPL-compatible

✅ Approved Libraries for Stem Separation

Option 1: PyTorch/LibTorch

• License: BSD-3-Clause
• ✅ GPL Compatible: Yes (BSD is GPL-compatible)
• Open Source: ✅ Full source available
• Repository: https://github.com/pytorch/pytorch
• Download: https://pytorch.org/cppdocs/installing.html

Option 2: ONNX Runtime (Recommended)

• License: MIT License
• ✅ GPL Compatible: Yes (MIT is GPL-compatible)
• Open Source: ✅ Full source available
• Repository: https://github.com/microsoft/onnxruntime
• Advantage: Smaller footprint, easier integration

Option 3: TensorFlow Lite C++

• License: Apache 2.0
• ✅ GPL Compatible: Yes (Apache 2.0 is GPL-compatible)
• Open Source: ✅ Full source available
• Repository: https://github.com/tensorflow/tensorflow

✅ Model Compatibility

Demucs Models

• License: MIT License
• ✅ GPL Compatible: Yes
• Open Source: ✅ Full source available
• Repository: https://github.com/facebookresearch/demucs
• Models: Can be converted to ONNX format

🚫 Libraries to AVOID

• Any proprietary/closed-source libraries
• Libraries with incompatible licenses (some commercial licenses)
• Virtual DJ SDK (proprietary)

📝 Recommendation

Use ONNX Runtime + Demucs ONNX models:

1. ✅ Fully GPL-compatible (MIT License)
2. ✅ Smaller footprint than LibTorch
3. ✅ Easier C++ integration
4. ✅ Official Demucs ONNX support available
5. ✅ Better for distribution

Implementation Plan

1. Download pre-trained Demucs ONNX models
2. Integrate ONNX Runtime C++ API
3. Create native C++ stem separator
4. Replace Python dependency completely

This approach ensures 100% license compliance while achieving our goal!