Our research effort is to pursue several different techniques for measuring physiological quantities from acoustic data. In one project, we are developing a “mouth sonar” apparatus for measuring the vocal tract shape. An externally generated acoustic signal is applied to the vocal tract at the lips, and the acoustic reflection is analyzed to determine the vocal tract area function.
Another project concerns the measurement of the pattern of vocal fold vibration occurring at the larynx. Linear Predictive Coding (LPC) analysis can be applied to the acoustic speech signal to determine an inverse filter. This filter serves to remove the influence of the vocal tract resonances to reveal the underlying train of airflow pulses produced by the opening and closing of the vocal folds. We are studying the use of parametric models of the glottal airflow pulses signal as a means of detecting modes of vocal fold vibration that are characteristic of the normal larynx and the larynx of individuals with voice disorders.
To conduct experimental studies on the acoustic signal recorded from speech, we have developed the CSpeech software package for digitizing, editing, and analyzing audio signals on the IBM Personal Computer. This software package is being used campus wide in the Departments of Electrical and Computer Engineering, Communicative Disorders, Otolaryngology, Linguistics, English, Communication Arts, and East Asian Languages and Literature. This common software package facilitates interdisciplinary projects in the study of speech.