A simplified vocal tract model explores the production of sibilant fricatives

The underlying physics of the vocal tract producing our precise speech sounds is a vital aspect of phonetics, linguistics and the treatment of speech difficulties. But simulating and analyzing these processes in a laboratory with suitably realistic accuracy is challenging. A research group reports in Physics of Fluids on a new simplified vocal tract model that accurately reveals the sound generation mechanisms of the (echomimetic) sibilant fricatives /s/ and /∫/.

The group built upon two earlier models: specifically, a computational model based on computed tomography images of a male Japanese subject producing the /s/ and /∫/ fricatives common to English-language consonants, and a realistic replica of the subject’s vocal tract created using a 3-D printer. While these models reproduced the sounds, they did not reveal detailed flow dynamics and their relation to sound production.

The new model was a simplified 3-D rectangular channel, constructed with acrylic boards simulating the vocal cavity, tongue, alveolar ridge plate, and teeth. A compressor delivered airflow at the subject’s measured flow rate and sound was measured by an omnidirectional microphone in a quiescent environment.

Based on the results, the simplified model accurately produced the flow field and spectral shapes of both /s/ and /∫/ sounds. The study showed that the characteristic peak frequency of each sibilant fricative depends on the tongue position and the consequent changes in jet flow and pressure downstream from the resulting vocal tract constriction. The next step of this work will be to confirm these findings using a realistic 3-D model.

Source: “Experimental and numerical investigation of the sound generation mechanisms of sibilant fricatives using a simplified vocal tract model,” by Tsukasa Yoshinaga, Kazunori Nozaki, and Shigeo Wada, Physics of Fluids (2018). The article can be accessed at https://doi.org/10.1063/1.5013632 .