Velopharyngeal insufficiency studied using finite-element models of male vocal tract with experimental verification

Finite element (FE) models of acoustic spaces corresponding to the human nasal and vocal tract for vowel /a/ are used for numerical simulations. Simplified FE model of the vocal tract for English vowel was created from geometrical data published in literature and for the Czech vowel by transferring data directly from MRI images. The nasal cavities were added to the models manually according to anatomical literature. The acoustic signal for the vowel / a / is simulated using transient analysis of the FE models in time domain. The vocal tract is excited by time dependent displacement of a small circular plate moving at the position of the vocal folds. The time response and frequency response functions are calculated near the lips, nostrils and at the vocal folds. Effects of velofaryngeal insufficiency are simulated and compared to results from acoustic measurements. In the previous papers of the authors [1,2] the acoustic frequency-modal characteristics of the human vocal tract were studied by FE modelling including the effects of cleft palate [3]. Here the study is extended to the time domain analysis using a real type of excitation of the acoustic spaces by pulses generated at the vocal folds. The simplified FE model of a male vocal tract for the English vowel /a/ was developed according to the MRI data published by Story et al. [4]. The FE model approximating the human supraglottal tract including the added nasal cavity spaces is presented in Fig. 1. The total length of the vocal tract from the vocal folds (on the right) to the lips (on the left) is 174.58 mm. The FE model used for simulation of phonation of the Czech vowel /a/ is shown in Fig. 2a [1]. A small connection (size of 20 finite elements) of the nasal and oral cavities was considered in the back area of the soft palate modelling the velofaryngeal insufficiency. The acoustic transient analysis was realised by the system ANSYS 5.7 using the acoustic finite elements FLUID30 considering the speed of sound c 0 = 353 ms-1 and the air density U 0 = 1.2 kgm-3. Zero acoustic pressure (p = 0) was assumed at the lips and nostrils. Other boundary walls of the acoustic spaces were considered to be acoustically absorptive. Fig. 1 FE model of male vocal tract for English vowel /a/ including the nasal cavity. The acoustic damping, which is associated with the fluid-structure interface …