3aED3. A transmission-line model of the vocal tract during fricative production.

Session: Wednesday Morning, May 15

Author: Jesse Ibarra
Location: Dept. of Elec. Eng., UCLA, 405 Hilgard Ave., Los Angeles, CA 90095


The purpose of this study is to examine the articulatory-to-acoustic mapping of the fricatives /s,sh,z,f,v,th/. The study utilizes MRI-derived vocal-tract area functions for two subjects [Narayanan et al., J. Acoust. Soc. Am. 98, 1325--1347 (1995)] and an articulatory synthesizer that is based on the analog circuit simulator HSPICE. The synthesizer [Rael et al., J. Acoust. Soc. Am. 97, 3430(A) (1995)] uses a transmission-line analog model of the vocal tract with the specification of the type and location of dependent or independent sources (voltage or current). Modeling involved placing noise sources in the vicinity of the maximum constriction and at other locations in the vocal tract. A voicing source(KLGLOTT88) was used for the voiced fricatives. Synthesized spectra matched naturally spoken spectra of /s,sh,v,z/ by placing two or three voltage sources near the constriction and near the lips for /s,z,v/, and at 1.2 cm away from the lips for /sh/. The current (monopole) source was not useful in synthesizing these fricatives. Results for /v,z/ were satisfactory using both voicing and noise sources even though /v/ was spectrally flat above 1.5 kHz. The model did not successfully synthesize /f,th/; this may be attributed to the seemingly more complex radiative and turbulent behavior for these sounds. [Work supported by NSF.]

from ASA 131st Meeting, Indianapolis, May 1996