This study analyzes the feasibility of recovering skin spectral characteristics from optical diffusion spectroscopy data using an analytical model. The model accounts for a semi-infinite medium geometry, a limited detector aperture, and the presence of a refractive index mismatch at the medium boundary. The employed approach involves two key stages. First, the spectral dependencies of the diffusion parameters the effective attenuation coefficient and the transport mean free path are recovered. Subsequently, the spectra of the absorption coefficient and the transport scattering coefficient are reconstructed from these parameters. The accuracy of recovering the spectral characteristics of homogeneous dermis is compared for two scenarios: a contact probe (external refractive index n_out=1.5) and a non-contact probe (n_out=1). The analysis is performed for various levels of blood volume content and different sets of source-detector distances (1.8, 2.8, 3.8, and 4.8 mm). The results show that the highest recovery accuracy (deviation not exceeding 10%) for both absorption and scattering spectra is achieved using the nearest three detectors in conjunction with contact probing. For non-contact registration using the same set of detectors, inaccuracies in recovering the transport mean free path in the near-infrared region lead to an overestimation of the absorption coefficient in this spectral range by up to 50%. Furthermore, the effect of a superficial layer (epidermis) on the recovery of the underlying layer's (dermis) spectral characteristics is analyzed. The presence of the epidermis has practically no impact on the accuracy of recovering dermis parameters when using contact probing. In contrast, the use of a non-contact probing leads to a systematic underestimation of the absorption coefficient and a systematic overestimation of the scattering coefficient. Keywords: diffuse reflectance spectroscopy, analytical model, Monte Carlo simulations.