Airborne Sound Transmission Modeling of Walls based on Random Point Process Theory
* Presenting author
A new method is presented to compute the diffuse sound transmission through walls that are either modeled deterministically or are taken to carry a diffuse vibration field. The fact that the natural frequencies can be considered to form a sequence of random points on the frequency axis makes this problem amenable to analysis by random point process theory. Furthermore, the method makes use of the fact that for any diffuse subsystem the statistics of its undamped eigenvalues and mode shapes saturate to universal distributions. The mode shapes saturate to Gaussian random fields, with a spatial covariance function depending only on wavelength and distance. The statistics of the local spacings between the eigenvalues saturate to those of a Gaussian orthogonal ensemble (GOE) matrix. This stochastic modelling approach is computationally very efficient as illustrated for a calcium silicate wall and a double glazing.