Some of these songs are _really_ good. And pretty far ahead of where I thought audio AI was. Does anyone know what models and tools are used? I imagine once you have stems you can maybe style / genre transfer and use a reference audio file perhaps? Like image and video models.