While considerable efforts have been expended to develop snore-driven markers for detecting obstructive sleep apnea (OSA), there is little emphasis on the relationship between the human upper airway (UA) dimensions and the attributes of snores. This paper aims to investigate the acoustical and perceptual impacts of changing the cross-sectional areas (CSA) of the pharynx and oral cavity on the production of snores. Synthetic snores were generated based on the source-filter theory, whereas natural snores were recorded from 40 snorers during nocturnal polysomnography. First formant frequency (F1), spectral peak frequency (PF), and psychoacoustic metrics (loudness, sharpness, roughness, fluctuation strength, and annoyance) of CSA perturbations were examined, completed with diagnostic appraisal of F1 and PF for single- and mixed-gender groupings using the receiver operating characteristic curve analysis. Results show that (1) narrowing the pharyngeal airway consistently increases F1, but not for PF; and (2) altering the airway dimensions yield no considerable differences in perception of snore sounds, but indirectly affect the psychoacoustics by changing the dynamics of snore source flow. Diagnostic outcomes for all groupings (p-value < 0.0001) demonstrate that F1 is more capable of distinguishing apneic and benign snorers than PF due to the close association of F1 with the UA anatomical structures. Correlation exists between the UA anatomy and the properties of snores; there is a promising future for developing snore-driven screening tools for OSA.