Date of Award

Spring 2015

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Biomedical Engineering

First Advisor

Johnson, Michael T.

Second Advisor

Berry, Jeffrey J.

Third Advisor

LaDisa, John

Abstract

Recent work with Electromagnetic Articulography (EMA) has shown it to be an excellent tool for characterizing speech kinematics. By tracking the position and orientation of sensors placed on the jaws, lips, teeth and tongue as they move in an electromagnetic field, information about movement and coordination of the articulators can be obtained with great time resolution. This technique has far-reaching applications for advancing fields related to speech articulation, including recognition, synthesis, motor learning, and clinical assessments. As more EMA data becomes widely available, a growing need exists for software that performs basic processing and analysis functions. The objective of this work is to create and demonstrate the use of new software tools that make full use of the information provided in EMA datasets, with a goal of maximizing the impact of EMA research. A new method for biteplate-correcting orientation data is presented, allowing orientation data to be used for articulatory analysis. Two examples of applications using orientation data are presented: a tool for jaw-angle measurement using a single EMA sensor, and a tongue interpolation tool based on three EMA sensors attached to the tongue. The results demonstrate that combined position and orientation data give a more complete picture of articulation than position data alone, and that orientation data should be incorporated in future work with EMA. A new standalone, GUI-based software tool is also presented for visualization of EMA data. It includes simultaneous real-time playback of kinematic and acoustic data, as well as basic analysis capabilities for both types of data. A comparison of the visualization tool to existing EMA software shows that it provides superior visualization and comparable analysis features to existing software. The tool will be included with the Marquette University EMA-MAE database to aid researchers working with this dataset.

COinS