Nowadays, porous absorption materials need to be environmentally sustainable. This concerns the sources of raw materials, low energy consumption in the manufacturing process, and options for recycling. Therefore, alternatives for mineral wools are needed and biomaterials offer an attractive solution. Some recently studied sustainable acoustic materials, such as foam-formed wood fibers, can be produced with negative carbon footprint. However, sound absorption performance must match the less sustainable variants to ensure competitiveness. Therefore, the prediction of sound absorption is important because it supports optimization of processes and functionality to achieve the best possible solution. This includes choosing the right model, taking the microstructure of the material into account. However, it has been observed that the microstructure of foam-formed pulps can change with increasing density, challenging the choice between foam and fiber-based models. In this study, foam-formed wood pulps with increasing density are produced, characterized and modeled by fibrous and foam-based analytical models. The model accuracy across densities is then evaluated against experimentally measured sound absorption. The results show that foam-based models become more precise at higher densities, while the fibrous model’s precision deteriorates. The finding contributes to the optimization of foam-formed pulps by choosing the proper model for the aimed density.