In Computational aeroacoustics, hybrid approaches first resolve the source and nearfield regions of the flow field by employing Reynolds Averaged Navier-Stokes (RANS) equations, Large Eddy Simulations (LES) or Direct Numerical Simulations (DNS).The source region data is used to form source terms, which are, in turn, applied to either empirical models or equations linearized around a mean flow. An acoustic analogy type of model is used to propagate the acoustics to the farfield regions. The aim of this research is twoold: to introduce and test a hybrid acoustic analogy, based on a coupling between the Navier-Stokes equations, applied in the source region, and the Non-linear Euler (NLE) equations applied in the acoustic propagation region; and to test and validate a recently derived generalized acoustic analogy theory in the framework of jet noise with acoustic source information obtained from RANS or LES. In the first part, the coupling between the Navier-Stokes and the NLE equations is accomplished via a buffer region, which is used to interpolate and penalize the flow variables of interest from the source region. The penalized flow variables are then applied as source terms in the NLE equations, to calculate the acoustic propagation. The non-linear Euler equations, discretized using highurate dispersion-relation preserving schemes constitute a very efficient approach for jet noise predictions in complex environments, especially for supersonic and hypersonic jets, where nonlinearities may propagate over long distances. In the second part, a RANS- or LES-informed model, which is used to provide data for Goldstein's generalized acoustic analogy, is presented. The generalized acoustic analogy of Goldstein is considered, wherein the effects of non-parallelism are taken into account and an asymptotic expansion is utilized to simplify the adjoint Green function equations. The use of the adjoint Green's function leads to a simple model for jet noise predictions for low frequencies and small observation angles, in the linear regime. Both approaches are extensively tested and validated against numerous benchmark problems and applications.