Change search
ReferencesLink to record
Permanent link

Direct link
Finite element generation of vowel sounds using dynamic complex three-dimensional vocal tracts
KTH, School of Computer Science and Communication (CSC), Speech, Music and Hearing, TMH.ORCID iD: 0000-0002-8991-1016
KTH, School of Computer Science and Communication (CSC), Speech, Music and Hearing, TMH.ORCID iD: 0000-0003-4532-014X
2016 (English)In: ICSV 2016 - 23rd International Congress on Sound and Vibration: From Ancient to Modern Acoustics, International Institute of Acoustics and Vibrations , 2016Conference paper (Refereed)
Abstract [en]

Three-dimensional (3D) numerical simulations of the vocal tract acoustics require very detailed vocal tract geometries in order to generate good quality vowel sounds. These geometries are typically obtained from Magnetic Resonance Imaging (MRI), from which a volumetric representation of the complex vocal tract shape is obtained. Static vowel sounds can then be generated using a finite element code, which simulates the propagation of acoustic waves through the vocal tract when a given train of glottal pulses is introduced at the glottal cross-section. A more challenging problem to solve is that of generating dynamic vowel sounds. On the one hand, the acoustic wave equation has to be solved in a computational domain with moving boundaries, which entails some numerical difficulties. On the other hand, the finite element meshes where acoustic wave propagation is computed have to move according to the dynamics of these very complex vocal tract shapes. In this work this problem is addressed. First, the acoustic wave equation in mixed form is expressed in an Arbitrary Lagrangian-Eulerian (ALE) framework to account for the vocal tract wall motion. This equation is numerically solved using a stabilized finite element approach. Second, the dynamic 3D vocal tract geometry is approximated by a finite set of cross-sections with complex shape. The time-evolution of these cross-sections is used to move the boundary nodes of the finite element meshes, while inner nodes are computed through diffusion. Some dynamic vowel sounds are presented as numerical examples.

Place, publisher, year, edition, pages
International Institute of Acoustics and Vibrations , 2016.
Keyword [en]
Acoustic wave propagation, Acoustic waves, Finite element method, Geometry, Linguistics, Magnetic levitation vehicles, Magnetic resonance imaging, Wave equations, Arbitrary Lagrangian Eulerian, Computational domains, Finite element codes, Finite element meshes, Stabilized finite element, Three dimensional (3D) numerical simulation, Vocal tract shape, Volumetric representation, Acoustics
National Category
Otorhinolaryngology General Language Studies and Linguistics Social Sciences Interdisciplinary Interaction Technologies
URN: urn:nbn:se:kth:diva-195461ScopusID: 2-s2.0-84987899372ISBN: 9789609922623OAI: diva2:1045810
23rd International Congress on Sound and Vibration, ICSV 2016, 10 July 2016 through 14 July 2016

QC 20161110

Available from: 2016-11-10 Created: 2016-11-03 Last updated: 2016-11-10Bibliographically approved

Open Access in DiVA

No full text


Search in DiVA

By author/editor
Dabbaghchian, SaeedEngwall, Olov
By organisation
Speech, Music and Hearing, TMH
OtorhinolaryngologyGeneral Language Studies and LinguisticsSocial Sciences InterdisciplinaryInteraction Technologies

Search outside of DiVA

GoogleGoogle Scholar

Total: 28 hits
ReferencesLink to record
Permanent link

Direct link