The Lattice Boltzmann Method (LBM) is an alternative technique for the simulation and mod- elling of fluid flows based on the Boltzmann equation. One of its advantages is its ability to handle very complex geometries with massively parallel computing. The LBM has achieved great success in simulating nearly incompressible and isothermal fluid flows, but it restricts its appli- cation range. A particularly active topic of investigation is its extension to more complex flows (e.g. multi-phase, thermal, compressible). We propose a pressure-based prediction-correction hybrid LBM model compatible with nearest-neighbor lattices (D2Q9 and D3Q19) with a single time relaxation process, to simulate subsonic and transonic compressible flows without shock. The approach is hybrid: mass and momentum conservation equations are computed using a LBM solver while an entropy conservation equation is solved via a finite difference approach. Following, an adequate forcing term is added to reproduce a correct viscous stress tensor and hybrid recursive regularized approach is used to stabilize the solution. Discretization of the entropy equation with viscous heat dissipation, in the finite difference part of the solver, is studied to improve accuracy of the scheme and to reduce the cost of calculations. Validation of this new method is carried out on a number of canonical cases, systematically challenging the coupling between velocity, pressure and temperature, including pressure wave propagation, and thermal Couette flows.