Conventional element based methods for modeling acoustic problems are limited to low-frequency applications due to the huge computational efforts.For high-frequency applications, probabilistic techniques,such as statistical energy analysis (SEA),are used.For mid-frequency range, currently no adequate and mature simulation methods exist.Recently,wave based method has been developed which is based on the indirect TREFFTZ approach and has shown to be able to tackle problems in the mid-frequency range.In contrast with the element based methods,no discretization is required.A sufficient,but not necessary,condition for convergence of this method is that the acoustic problem domain is convex.Non-convex domains have to be partitioned into a number of (convex) subdomains.At the interfaces between subdomains,specific coupling conditions have to be imposed. The considered two-dimensional coupled vibro-acoustic problem illustrates the beneficial conver- gence rate of the proposed wave based prediction technique with high accuracy.The results show the new technique can be applied up to much higher frequencies. Conventional element based methods for modeling acoustic problems are limited to low-frequency applications due to the huge computational efforts. For high-frequency applications, probabilistic techniques, such as statistical energy analysis (SEA), are used. For mid-frequency range, no adequate and mature simulation methods exist. Really, wave based method has been developed which is based on the indirect TREFFTZ approach and has shown be able to tackle problems in the mid-frequency range. In contrast with the element based methods, no discretization is required. A sufficient, but not necessary, condition for convergence of this method is that the acoustic problem domain is convex. Non-convex domains have to be partitioned into a number of (convex) subdomains. At the interfaces between subdomains, specific coupling conditions considered to be imposed. The considered two-dimensional coupled vibro-acoustic problem shows the beneficial convergence rate of the proposed wave-based predictive tion technique with high accuracy. The results show the new technique can be applied up to much higher frequencies.