Acoustic tube model of the human vocal tract: formants and vowels

Abstract

The identity of vowels in speech and singing is determined by the eigenmodes of the human vocal tract (VT); in particular the two lowest modes—which give rise to the so-called formants—are essential. We present a two-tube model to analyse the influence of VT geometry on the formants. It models jaw opening and tongue position in the human VT. The analysis is based on the dispersion relation of plane waves and uses only elementary concepts in physical acoustics. Our model agrees reasonably well with measured data on formants for the centre, [ə], and the corners, [a:], [i:] and [u:], of the vowel triangle. The limitations of the model and the possible extension to a three-tube model are outlined.