Extract

Abstract

A new method for the reconstruction of pseudo-3D tongue surfaces in sustained speech sounds is presented. A head and transducer stabilizer was used to acquire series of coronal ultrasound images in three participants (2 females, 1 male). The English speech sounds /a/, /i/, /u/, /n/ and /ŋ/ were recorded. The tongue height information from the coronal ultrasound images was used to reconstruct pseudo-3D surfaces of the tongue. Results demonstrated an acceptable intra-trial variability. An anteriority measure provided information on the protrusion of the tongue during different speech sounds. A concavity measure summarized information about the grooving of the tongue. Differences between coronal and sagittal ultrasound images of the tongue in relation to pseudo-3D imaging are discussed.

Keywords

Tongue, ultrasound, 3D imaging, speech

*   Corresponding author: tim.bressmann@utoronto.ca, Tel: +1 416 978 7088, Fax: +1 416 978 1596

a   Department of Speech-Language Pathology, University of Toronto, 160–500 University Avenue, Toronto, Ontario M5G 1V7, Canada ← 171 | 172 →

1.0   Introduction

There are multiple instrumental methods for obtaining data about tongue shapes. However, imaging methods such as magnetic resonance, videofluoroscopy or ultrasound tend to be applied exlusively to the midsagittal plane. While many relevant aspects of information can indeed be obtained from the midsagittal plane, the tongue is a complex three-dimensional muscular structure, and restricting the data acquisition to the midsagittal plane results in a potential loss of information. There are different applications where the 3D-shape...