Recent Advances in Multi-Paradigm Physical Modeling

Physics-based modeling of sound sources, such as of musical instruments and speech production, is based on several different modeling paradigms or methodologies. These can be roughly categorized as finite difference schemes, mass-spring models, modal decomposition methods, digital waveguide models, wave digital filter models, and source-filter models. The three first methods are based on the use of so-called K-quantities, while the next two ones use wave quantities for computation, and the last one is a category between true physical modeling and signal processing algorithms. Research groups in physics-based modeling have traditionally applied only their one or two favorite paradigms. Recently there has been growing interest to utilize any or many of the approaches, according to specific needs. This includes also combining the paradigms to hybrid modeling in order to optimize the performance or flexibility of design. The trend has improved the overall understanding of physics-based modeling, and will probably lead to many new models and ideas, as well as to tools that support multi-paradigm approaches. This paper presents an overview of multi-paradigm techniques in physic-based modeling, as applied particularly to musical instruments. It will also discuss some recent work in the Sound Source Modeling Group at the Helsinki University of Technology.