The reverberation chamber method to measure a material’s absorption coefficient has known limitations. In particular, the measured value can exceed unity, due to the finite size of the studied sample, which creates edge diffraction effects. These are not accounted for in classical theories of absorption. A revised theoretical approach is proposed to evaluate the absorption of sound by a finite-sized absorber flush-mounted in an infinite rigid baffle. The absorption, reflection and transmission of sound through the sample are governed by the material’s surface impedance (dissipation of energy) and the sample’s radiation impedance, which depends on its geometry. The maximum possible absorption is achieved in an “open-window” configuration, in which sound passes freely through an aperture of the same dimensions as the sample. In that case, the absorbed sound comprises entirely
of the sound transmitted to the other side of the baffle. For an open window, the absorption coefficient drops at low frequencies and at large incidence angles, due to edge diffraction. Therefore, we propose to compare sound absorption by a given sample to the corresponding open-window configuration. Comparisons with experimental data show that the revised theory makes it possible to better predict the absorption coefficient measured in a reverberation chamber.