Acoustic modeling of fricative /s/ for an oral tract with rectangular cross-sections

Abstract

Fricative /s/ is known to be pronounced by jet generation and subsequent impact on walls of the oral cavity. The prediction of acoustic characteristics of /s/ is an ongoing research topic due to the aeroacoustic nature of the sound source. In this study, acoustic characteristics are modeled using the multimodal theory with monopole and dipole sources positioned in the oral tract waveguide. The oral tract geometry of fricative /s/ was simplified by concatenating rectangular channels whose cross-sectional areas and vertical height are derived from medical imaging. To validate the model accuracy, transverse and sagittal directivity patterns (49 cm every 15°) were measured for flow supplied to a realistic oral tract replica. Comparison between measured and modeled spectra showed that the modeling with the dipole source predicted the pressure amplitude within a discrepancy of ±5 dB up to 14 kHz. Modeled acoustic directivity patterns using a dipole source reflected main tendencies observed on measured directivity patterns in both the transverse and sagittal planes. The proposed modeling approach enables a systematic analysis of the high frequency (>5 kHz) acoustic characteristics as a function of geometrical details for speech production due to an aeroacoustic sound source.