Vibroacoustical systems as well as their behavior are coupled by nature. However, it may be of first interest for engineers to work on decoupled models, so that behavior of the structural/acoustic sub-system can be easily predicted and investigated. This work focuses on a virtual decoupling approach for vibroacoustics, with the objective to reconstruct decoupled and reduced system matrices from coupled experimental measurements. As a promising identification technique, the properness enforcement method has been developed in the literature for both symmetric and non-symmetric systems. During the decoupling process, however, this methodology still needs to be improved from two aspects: 1) the introduction of an additional correlation process so that the structural/acoustic sub-model can be correctly extracted from the coupled system; and 2) the additional optimization step after the complex vectors are approximately corrected by the properness enforcement method. These two key points are addressed by an integrated framework containing three aspects (i.e. identification, optimization, and correlation), which are specially designed for vibroacoustical applications. The finally identified system matrices of a decoupled and reduced equivalent system can exhibit the same behavior as the experimentally measured one. A simulated example is first presented to illustrate the use of this approach in detail. Then an experimental case study is used to demonstrate its feasibility in engineering applications.