A preliminary acoustic analysis of three-dimensional shape of the human nasal cavity and paranasal sinuses extracted from cone-beam CT

The shape of the nasal cavity and paranasal sinuses is complex and varies between individuals. Because the shape is almost stable during speech, the acoustic properties could constantly provide speaker specific information to speech sounds, that is, speaker individuality. In this preliminary analysis, the shape was extracted from cone-beam CT data for a subject using a machine learning technique and its acoustic properties were examined using finite-difference time-domain simulation. The transfer function from the glottis to the nostrils was calculated and the distribution pattern of the pressure anti-nodes was visualized at frequencies of major spectral peaks and dips. In addition, transfer functions were calculated when each of the paranasal sinuses other than the ethmoidal ones was occluded to identify which sinus caused which dip. As a result, the longitudinal resonance in the right or left half of the nasal cavity generated each peak, while the transverse resonance in the pharyngeal cavity caused the major dips. The right maxillary, sphenoidal, and frontal sinuses generated dips below 1 kHz. The left maxillary sinus and ethmoidal sinuses, however, contributed to generating not only dips but also peaks. [This research was partly supported by JSPS KAKENHI Grant Numbers 15K00263, 25280066, and 25240026.]