Two-Dimensional Symmetric Digital Waveguide Model for the Vocal Tract

Abstract

The existing waveguide modeling of the vocal tract has significant limitations concerning acoustic accuracy and simulation cost. The multi-dimensional waveguide modeling gives precise acoustic wave propagation in the vocal tract at the expense of high computational cost. The efficiency of multi-dimensional digital waveguide modeling has remained a big concern. In this work, we evaluate a new implementation that intends to overcome these limitations. In this paper, we examine a new implementation that aims to address this issue. We introduce the symmetric two-dimensional digital waveguide model by taking half of the mesh with the symmetric line. The symmetric conditions are implemented on the symmetric line on the meshing of the digital waveguide model of the vocal tract. For the evaluation of the current model, the standard two-dimensional non-symmetric digital waveguide model is taken as a benchmark model. The comparison of our symmetric model with the benchmark model has been presented in terms of frequency profiles, formant frequencies, and efficiency. The phonetic vowels /æ/, /ʌ/, /I/, /ɝ/, and /o/ have been chosen for the demonstration of the proposed work. In the simulation of the current work, the symmetric model is found to be about identical to that of the benchmark model for frequency profiles and formant frequencies. In all cases taken in the proposed work, the symmetric digital waveguide model is more than 90% more efficient than the benchmark model.