State space model of airflow in the human vocal apparatus

To simulate both airflow and air pressure required for speech production, we developed a nonlinear state-space model of airflow in the human speech apparatus. We modeled the lungs as a mechanical, force-driven piston pump venting into a simplified larynx model, represented as a valve with time-varying resistance to airflow. The pump incorporates the effects of elasticity, viscosity, friction, and inertia, as well as differential air pressure. To maintain a constant target airflow through the larynx, a proportional-derivative (PD) controller applies force and regulates the piston. The model was implemented in MATLAB. Simulation results demonstrate that the model can maintain a constant airflow suitable for speech production while compensating for varying airflow resistance. This study lays the groundwork for developing a mechanical lung that can be integrated into a robotic vocal system with realistic airflow characteristics