Detailseite
Projekt
Determination of boundary conditions by inverse modeling of room acoustics
Antragsteller
Professor Dr. Michael Vorländer
Fachliche Zuordnung
Akustik
Förderung
Förderung von 2011 bis 2014
Projektkennung
Deutsche Forschungsgemeinschaft (DFG) - Projektnummer 174776315
During investigations in the first phase of this subproject it was observed that major uncertainties in room acoustics simulations were due to the user input rather than to the algorithm cores. This applied mainly to absorption and scattering coefficients. This makes it crucial to apply absorption coefficients with reasonable care. Unfortunately, the necessary accuracy is often not reached with common procedures. In many cases it was observed that this data cannot directly be used in room acoustics simulations with reliable success. Therefore simulations are almost always re-'tuned' - at least if reference measurements are available. In general it must be said that the possibilities of computer simulations are limited by the available quality of input data - which often is the weakest element of the simulation chain.The applicant proposed and started developing a new method for the acquisition of absorption coefficients in-situ. It accounts for the intended use in computer simulations by reversely applying the simulation model for the calculation of absorption coefficients. Typical room impulse response measurements serve as an input for an inverse room acoustics model.The inversion of the room acoustics model in the simulation framework RAVEN is therefore in the focus of the application for the second phase. This is possible if a matrix representation of the sound propagation in the room can be found. Using trajectories of sound rays in the geometrical simulation model such a matrix was successfully formulated. The inverse was constructed using an iterative numerical approach using the optimization toolbox of MATLAB. A non-linear least squares method with automatic differentiation was applied.The first main goal of subproject P2 is to continue und finish the development of the inverse approach. This includes the determination of requirements and limits of the method as well as algorithmic improvements. Especially in the optimization process a lot of potential can be expected from advances in the objective function. This includes additional cost functions that reflect room acoustical considerations as well as full formulations of the first and second order derivatives to replace the automatic differentiation. Using measurements from controlled test environments the inverse method will be validated and its applicability for the acquisition of absorption coefficients will be assessed.The second main goal of subproject P2 is the establishment of key data and of reference information for the international round robin on auralization. For this purpose the acoustic data of room surfaces must be determined, which is done using the presented inverse method and additional control measurements using the in-situ pressure-velocity-probe method. The reference input data (room geometry, surface properties, source and receiver characteristics) is then provided to the participating simulation programs together with guidelines for an optional 'tuning' step.
DFG-Verfahren
Forschungsgruppen
