Modeling of a speech production system based on MRI measurement of three-dimensional vocal tract shapes during fricative consonant phonation

This study, based on the measurement of three-dimensional (3-D) vocal tract shapes during fricative consonant phonation, presents a realistic modeling of a human speech production system. The 3-D shapes of a vocal tract and a dental crown were measured using Magnetic Resonance Imaging (MRI). A male subject was asked to produce the fricatives /s/ and /6/ while wearing a dental crown plate that contained a contrast medium for MRI processing. 3-D MR images of the vocal tract for each sound were obtained while the subject’s tongue was kept still. The 3-D shapes and area functions of the vocal tract corresponding to respective sounds were computed using a gray level interpolation technique to form serial sections. The measured results suggest that there are individual differences in speech production and vocal tract shapes. The airflow involved in the production of the fricatives /s/ and /6/ was estimated in the 3-D vocal tract using the Finite Element Method (FEM). The shapes of the 3-D vocal tracts for the fricatives were reconstructed from the coronal MR images. The behavior of the airflow was determined from the vector diagram of the flow rate. In this study, the vocal tract model with cascading circular tubes is called the VT model. A new acoustic model for the phonation of fricatives was proposed based on the VT model in which the sound source was a noise. Synthesized sounds of the Japanese fricatives /s/ and /6/ were generated using this model. An auditory test demonstrated that the generated sounds were intelligible.