David Roze is a researcher in the S3AM team. He will be presenting his Habilitation to supervise research entitled "Sound synthesis using physical models: non-linear dynamics and interactions for virtual instrument making with guaranteed passivity".
The defence will take place on Wednesday 11 December in the Salle Stravinsky at 9am.
It will be broadcast on Youtube :
The jury will be composed by :
M. Paul Kotyczka, Privatdozent Dr.-Ing. habil., TU München, Rapporteur
Claude-Henri Lamarque, Professeur des Universités, ENTPE-Université de Lyon, Rapporteur
Loïc Le Marrec, Professeur des Universités, Université de Rennes 1, Rapporteur
M. Kerem Ege, Maître de Conférences HDR, INSA de Lyon, Examinateur
M. Anthony Gravouil, Professeur des Universités, INSA de Lyon, Examinateur
M. Jean-Loïc Le Carrou, Professeur des Universités, Sorbonne Université, Examinateur
Mme Emanuelle Rouhaud, Professeure des Universités, UTT, Examinatrice
Abstract:
This habilitation presents the main results of my research work on sound synthesis using physical models. With the aim of improving sound realism, this project focuses mainly on simulating the dynamics of nonlinear resonators and their interactions.
This approach requires research into the representation and resolution methods best suited to three objectives:
- to model and compute the interaction between different resonators, so that these models can be used in virtual instrument making
- ensure simulation stability
- reduce computing time for real-time synthesis.
First of all, within the framework of the Volterra series spatio-temporal kernels have been defined in order to write interactions (based on the kernels of the inverse system) that are variable in time and space. The resulting structure makes it possible to simulate weakly nonlinear systems and their interactions at low cost.
More recently, the Port Hamiltonian Systems (PHS) formalism has been used to establish numerical methods guaranteeing a discretized power balance for linear and non-linear multi-physics systems and their interactions. This approach has also been used to establish control laws aimed at modifying the frequency, damping or non-linear behavior of real systems.