Automatic modeling and authoring of nonlinear interactions between acoustic objects

The physical synthesis of musical sounds is based on the modeling of the nonlinear interaction between two separate structures (at least one of which with desirable resonating properties). Such structures and their interaction are usually modeled with a global approach (as a single system), but flexibility and scalability issues suggest us that it would be highly desirable to adopt a block-wise approach, based on the individual synthesis and discretization of the interacting parts. In this paper we address the problem of how to automate the modeling of this nonlinear interaction and how to enable the on-the-fly fusion of model that corresponds to this interaction. We show how a solution based on binary connection trees can be fruitfully employed in an integrated modeling system that is able to automate this sound synthesis approach. We also show that, with this solution, the modeling of physical interactions can be done in an entirely symbolic (graphical) fashion. We finally provide a description of the graphical user interface for a user-friendly authoring of interactional models and an overview of a live performance system based on this technology.