kth_logo.gif

Internal Report

Modelling the Production and Propagation of Sound in Individual Human Vocal Tracts

Author Document Type Year Download File size
Lukas Schickhofer Doctoral thesis 2019 Download 17755 kB
Id
ISSN 0348-467X
ISRN KTH/MEK/TR--19/02--SE

Abstract

Voice generation and the expression through speech are of vital importance for communication. The human upper airways are the origin of the process of speech production, which involves a modulation of the periodically pulsed pressure from the lungs by the vocal tract volume. In this work, phonation and voiced speech are investigated through both low- and high-order models, which are applied to vocal tract geometries of increasing complexity. Initially, the effect of variations of vocal fold closure, fundamental frequency, and vocal tract length on the computed acoustic signal is examined through parameter studies based on one-dimensional wave reflection analogues. Eventually, unsteady large eddy simulations based on the compressible Navier-Stokes equations are carried out to compute the pressure fluctuations and the associated distribution of resonance modes as a result of the interaction with the static vocal tract. Thus it is possible to calculate tonalities from the entire audible range of frequencies from 20 to 20000 Hz. In particular the inharmonic broadband sound component produced predominantly by coherent structures in the upper airways and at frequencies above 2 kHz is resolved in the current study, which is not captured by low-order models based on wave equations. Furthermore, three-dimensional numerical meshes based on surface representations of the human upper airways under voiced speech from magnetic resonance imaging (MRI) data of a healthy male subject are applied. These are necessary to resolve high-order acoustic modes that would not be represented by simplified geometries. Validation and verification of the chosen methods are achieved through comparison with experimentally obtained speech data, as well as Helmholtz eigenfrequencies of the considered vowel pronunciations. The main scope of this work is the assessment of acoustic sources and the conditions for aerodynamic sound being produced and propagated in the upper airways during phonation. The distribution of acoustic sources involved in the generation of the dominant frequencies are identified by application of acoustic analogies as well as surface Fourier transformation of the acoustic pressure fluctuations. However, the human upper airways do not only embrace the source of phonation and affect the modulation of the voice. Moreover, unwanted sounds may be generated in the upper airways due to elastic, collapsible parts that are susceptible to flow-induced vibration and resonance. The sound resulting from fluid-structure interaction in the upper respiratory tract, commonly known as snoring, can be an important indicator for underlying breathing disorders, such as obstructive sleep apnea (OSA). In a smaller part of this project, the flow structures and acoustic sources as a result of the interaction of shear flow of various Reynolds numbers with an elastic element are computed. The geometric dimensions are chosen to be representative of average physical values of the upper respiratory tract. Onset of tissue vibrations and resonance effects are investigated for a range of parameters of both solid and fluid. The obtained results of this work are aimed to contribute also to the development of a computational tool that assists physicians in the assessment of the airway function and the effectiveness of treatment plans prior to their application.