One proposes a novel approach to model sound-insulation layers based on the use of the fuzzy structure theory which presents the advantage to account for variability as well as no addition of DOF in Computational mode Is. The keypoint of the method is the construction of a mean elastoacoustic sound-insulation layer model whose parameters are simply the modal density, the coefficient of participating mass and the damping coefficient. In this paper, the detailed construction of the mean model and an identification methodology of the mean parameters are presented. The modal density is identified from a stand-alone detailed finite element model of the sound-insulation layer. The coefficient of participating mass is then obtained by solving an inverse problem which is formulated as an optimization problem. Finally, the theory is validated with experiments for a vibroacoustic system made up of a steel plate connected to an elastic framework on its edges, covered with a sound-insulation layer and coupled with a bounded acoustic cavity.