Date of Award

Fall 2016

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Electrical and Computer Engineering

First Advisor

Johnson, Michael T.

Second Advisor

Berry, Jeffrey

Third Advisor

Povinelli, Richard

Abstract

Electromagnetic Articulography (EMA) has become an integral tool for researchers and clinicians who seek to characterize speech kinematics. The position and orientation of the articulators – which include the jaw, lips, and tongue – are recorded by attaching sensors to the articulators and tracking the movement of the sensors through an electromagnetic field. This has been used by researchers and clinicians to better understand dysarthria and synthesize speech, among other applications. Another speech tool, electroglottography (EGG), is used to analyze the movement of the vocal folds during speech production. This is achieved by measuring the time variation of the contact of the vocal folds and analyzing it with regards to the speech produced. Clinically, EGG is used to identify voice abnormalities, including those without visual or acoustic abnormalities. These systems are not used concurrently because of the electromagnetic field used with the EMA system; NDI and Carstens affirm that metal should be kept out of the field during EMA use. Concurrent use of these systems would lead to simultaneous measurements of the laryngeal and upper airway-articulatory abilities, which could increase understanding of motor speech issues. Parameters derived from the EGG signal could also be incorporated with articulatory parameters to improve synthesized voice quality and synthesize a more realistic voice. The objective of this research is to investigate whether the interference present between the EMA and EGG systems is significant and, if so, to characterize it so the systems can be used simultaneously. Analysis of the interference was obtained through several data collections. The first assessed the degree of interference when the EMA sensors were stationary. The second data set was collected using a model that maintained sensor orientation while the sensors were at a nonzero speed. The final data set was obtained with a model that in which the sensors were in a fixed position while changing the orientation of the sensors, with minimal translational velocity. Sources of interference that were present included the EGG system and orthodontia. The resulting data led to the conclusion that the presence of the EGG or the orthodontic appliances does not cause significant interference.

COinS