Dynamical model parameters from ultrasound tongue kinematics

Area of Science:

• Speech science
• Biomechanics
• Dynamical systems modeling

Background:

• Speech control is often modeled as a dynamical system where articulators move towards target positions.
• Traditional evaluation methods rely on fleshpoint data like electromagnetic articulography (EMA).
• Recent advances in ultrasound imaging offer a potential alternative for capturing articulatory movements.

Purpose of the Study:

• To assess if ultrasound tongue kinematics can reliably estimate parameters of a linear harmonic oscillator model.
• To compare these ultrasound-derived parameters with those obtained from simultaneously recorded EMA data.
• To evaluate the utility of ultrasound in assessing dynamical articulatory models.

Main Methods:

• Utilizing ultrasound tongue imaging to capture articulatory kinematics.
• Employing electromagnetic articulography (EMA) for simultaneous recording of articulatory data.
• Applying linear harmonic oscillator models to estimate dynamical parameters from both ultrasound and EMA data.
• Incorporating mandibular short tendon tracking for jaw motion analysis.

Main Results:

• Ultrasound and EMA yield comparable dynamical parameters for articulatory control models.
• Mandibular short tendon tracking effectively captures jaw motion.
• The study demonstrates the feasibility of using ultrasound for dynamical articulatory modeling.

Conclusions:

• Ultrasound kinematics are a reliable source for estimating dynamical parameters in speech production.
• The findings support the use of ultrasound imaging as a valid alternative to EMA for evaluating articulatory dynamics.
• This research advances the application of ultrasound technology in speech science and biomechanical modeling.