|
Numerical simulations of Japanese sibilant [s]| old_uid | 77 |
|---|
| title | Numerical simulations of Japanese sibilant [s] |
|---|
| start_date | 2005/10/17 |
|---|
| schedule | 13h30 |
|---|
| online | no |
|---|
| location_info | /, salle de réunion R |
|---|
| summary | Background Among the many speech problems caused by dental protheses1), the most common one generally occurs in newly fabricated dentures. By changing the user’s oral morphology, users can experience difficulties while articulating certain phoneme produced in the frontal area of the oral tract, such as sibilants [s], [z], [sh]. The fabrication of prostheses often includes the anterior teeth and palate. If the critical part in the oral cavity that generates a sibilant can be detected and incorporated in the fabrication, prostheses can be made to prevent possible speech difficulties for patients. A research shows that the sibilant [s] is produced by turbulences between the teeth2). However, currently no study has been done on the detection or visualization of its source. If the source can be detected and visualized, dentists will be able to recognize the critical part in the oral tract that affects the production of the sibilant [s]. On the other hand, many studies have been done on aero-acoustics, such as those performed by Lighthill3) and Powell4). In particular, Lighthill’s equation3) is the key to predict sound propagation at observation points. Lighthill’s acoustic analogy shows that the Lighthill equation can be applied to everything related to a sound. Under the isentropic condition, the Lighthill’s sound source term can be replaced by Powell’s sound source, which is why Powell sound source is determined in this study. Objective In this study, we propose a new method of speech analysis by numerical simulations, such as computational fluid dynamics (CFD) and computational aero-acoustics (CAA). Consequently, we were able to determine the sound source of the sibilant [s] by these methods. Material and Methods An oral tract model was obtained using a dental impression paste. During CFD application, a commercial software called "STREAM" (Software CRADLE, Japan) was used to determine the oral air flow when a subject articulated the sibilant [s]. The CFD solver only uses Reynolds Averaged Navier-Stokes equations model (RANS). The oral tract data was imported into the pre-software STREAM and then a computational model was constructed. As a result, 1,700,000 mesh as finite volume elements were applied. The calculation was performed by the supercomputer SX5 (NEC Corp., Japan), which has 2.4GB memory and takes 72 hours by 10GFLOPS. The CAA application served two purposes, one was to detect the sound sources of the sibilant [s] by Powell sound source, and the other was to determine sound propagation at the observation point. In this study, the former was carried out first. Powell sound source is described as ()divu£l£s°— £s and u are given by the result of CFD calculation. Results The oral air flow during the articulation of the sibilant [s] could be visualized, as shown in Figure 1. It confirmed that the highest velocity of the flow occurs between the anterior teeth. Powell sound source was visualized, as shown in Figure 2 and Figure 3. The result showed that the sound source is located around the anterior teeth. This meant the production of sibilant [s] is influenced by the shape and location of the anterior teeth, as well as the relationship between the maxilliary teeth and the mandibular teeth. |
|---|
| responsibles | Lécuyer |
|---|
| |
|