models that may be interesting for music use cases