A biomechanical model of cardinal vowel production: Muscle activations and the impact of gravity on tongue positioning

Abstract

A three-dimensional (3D) biomechanical model of the tongue and the oral cavity, controlled by a functional model of muscle force generation (lambda-model of the equilibrium point hypothesis) and coupled with an acoustic model, was exploited to study the activation of the tongue and mouth floor muscles during the production of French cardinal vowels. The selection of the motor commands to control the tongue and the mouth floor muscles was based on literature data, such as electromyographic, electropalatographic, and cineradiographic data. The tongue shapes were also compared to data obtained from the speaker used to build the model. 3D modeling offered the opportunity to investigate the role of the transversalis, in particular, its involvement in the production of high front vowels. It was found, with this model, to be indirect via reflex mechanisms due to the activation of surrounding muscles, not voluntary. For vowel /i/, local motor command variations for the main tongue muscles revealed a non-negligible modification of the alveolar groove in contradiction to the saturation effect hypothesis, due to the role of the anterior genioglossus. Finally, the impact of subject position (supine or upright) on the production of French cardinal vowels was explored and found to be negligible.

Limb versus Speech Motor Control: A Conceptual Review

“…Stone et al (2007) observed similar differences in tongue movements during speech production in upright versus supine position. Simulations based on a realistic 3D biomechanical tongue model that employed the same motor commands for upright and supine position reproduced these differences (Buchaillard et al, 2009). All these observations suggest that for both limb and speech movements, the motor system does not adjust motor commands to create movements that are invariant against changes in orientation relative to gravity, but endows movement with a sufficient amount of stability to remain functional under such varied conditions.…”

“…This has made it possible to develop realistic, complex, biomechanical models of the articulators (Wilhelms-Tricarico, 1995;Payan & Perrier, 1997;Dang & Honda, 2004;Gerard et al, 2006;Buchaillard et al, 2009). It has been shown, for instance, that trajectories of certain sounds are largely influenced by muscle anatomy and tongue-palate interactions (Perrier et al, 2003), as well as by fluid-soft tissues interactions (Perrier et al, 2000), and that velocity profiles can be determined by muscle fibers orientations (Payan & Perrier, 1997).…”

A Biomechanical Modeling Study of the Effects of the Orbicularis Oris Muscle and Jaw Posture on Lip Shape

“…The model had been previously adapted to the speaker's CT data using a segmentation of the interior bone surface and the exterior skin surface (Bucki et al, 2010). The reference jawtongue-hyoid bone model (Stavness et al, 2011) combined and registered two reference models to the same CT dataset: a 3D rigid-body jaw-hyoid bone model (Hannam et al, 2008) and a 3D FE tongue model (Gerard et al, 2006;Buchaillard et al, 2009). …”

Synthesis of VV Utterances from Muscle Activation to Sound with a 3D Model

“…Third, constructing the vocal tract 3D geometry is not as simple as calculating the area function in an articulatory model and especially not in a biomechanical model [12]. The first and second reasons are not valid anymore, since 3D articulatory biomechanical models [10,11] and acoustic models [13,14] are currently available. However, construction of a 3D vocal tract geometry, which is a requirement for the acoustic simulation, is still a challenge in particular when using articulatory models that include separate articulators.…”

“…An alternative to these approaches are methods using articulatory models, which can be either geometrical [6,7,8] or biomechanical [9,10,11]. Geometrical models are developed to replicate the kinematic of the articulation rather than its dynamics, while biomechanical models are more suitable for reproducing the speech dynamics.…”