Vegetal wools are highly porous materials made of vegetal and polymeric bicomponent fibers. They feature high-level multifunctional properties, such as sound absorption and hygrothermal regulation. However, despite their attractive properties, they suffer from poor low-frequency sound absorption for thin panels (less than 5 cm thick). A solution to this problem can be found in meta-material methods. Unfortunately, very little research has been done on this subject with applications for bio-based materials. Moreover, it seems relevant to use the specificities of vegetal wools, such as the polydispersity of their fiber radii, for these acoustic optimization approaches involving meta-material methods. To meet this challenge, a state-of-the-art has been carried out to identify suitable analytical modeling methods to simulate their acoustical behavior, and to optimize it. A first optimization consisted in implementing the meta-material concept of double porosity into a resistive hemp wool layer. Combined with a multiple layers configuration, this approach significantly improved the acoustic absorption of hemp wool panels. To go further, this work aims at comparing different analytical approaches to account for fiber radius polydispersity. Vegetal wools present a large range of fiber radii, and most studies consider a mean radius for simulations, with arithmetic or quadratic averaging methods. However, these approaches may overestimate the role of fibers with larger radii, which, for the example of static airflow resistivity relying on the thinner fibers, can inaccurately represent the physics. This work investigates different ways of averaging the radius probability distribution of vegetal fibers to improve the accuracy of analytical estimations.