The acoustic modeling of materials is of paramount importance for the accurate prediction of untested configurations and the development of optimal manufacturing strategies. Nevertheless, this task presents significant challenges due to the complexity of the parameters governing materials, such as porous media, each of which demands distinct experimental setups and procedures, often difficult to implement. Inverse methods for parameter estimation rely on physical approximations and require experimental protocols that are highly sensitive to boundary conditions, rendering the process both expensive and time-consuming. This study investigates the efficacy of machine learning techniques, particularly artificial neural networks, in determining the parameters of the Johnson-Champoux-Allard (JCA) model for porous samples, with a specific emphasis on the mutual influence of input features on network performance. Building upon these insights, a hierarchical artificial neural network-based procedure is proposed and validated using experimental data to predict the JCA parameters with minimal error.