A Thermoacoustically-Driven Vocal Tract Model

Abstract

Over centuries, different mechanical methods have been used to mimic the sounds of the human vocal tract. The researchers in this paper have used a thermoacoustic engine to produce vowel formant frequencies based on Arai’s three-tube sliding model [1]. The sliding constriction represents the tongue, the open end represents the mouth, and the closed end represents the glottis. A one-dimensional wave equation is used to calculate the formant frequencies as a function of tongue position. By analyzing the peak frequencies of the FFT spectrum from the engine, the more dominant modes were compared to the mathematical model to evaluate its predictive power. These results may be used in speech therapy and diagnosis.