Subglottal Resonances

I am interested in discovering the acoustical structure of the lower (subglottal) airway. This topic has received some attention from speech researchers, beginning with van den Berg’s seminal study in 1960, and from respiratory physiologists. The respiratory physiologists have been interested in the acoustics of the lower airway as a means for detecting lower airway pathologies, and the speech scientists have been interested primarily in the effects of subglottal resonances on vocal fold vibration. It has largely been assumed that the role of subglottal resonances in speech is negligible, given the relatively large impedance of the glottis.

While it is true that the impedance of the glottis is large, and therefore the effects of subglottal resonances in speech are small, they appear not to be entirely negligible. My research aims to discover the specific acoustic structure of the lower airway, its specific acoustic effects on speech, primarily in vowels, and the impact of these effects on the perception of speech. To this end, I am carrying out three kinds of experiments: 1) I have implemented a model of the vocal tract, including a detailed model of the lower airway, using MATLAB, and am exploring the theoretical acoustic properties of the system for different vowels; 2) I am analyzing speech acoustic data in combination with (indirect) accelerometer recordings of the lower airway acoustics; 3) I am carrying out speech perception experiments in which certain parameters of the lower airway acoustic properties are varied.

There are two important precedents to this work. In 2004, Chi and Sonderegger reported on a speech production study in which they correlated speech acoustic data (second formant, subglottal pole-zero pair) with accelerometer recordings (subglottal formants). The pole-zero pairs in the speech signal are derived from the formants of the subglottal system. They found that the second formant is higher in frequency than the second subglottal resonance in front vowels, and lower in back vowels. This finding corroborates the observation made in Ken Stevens’ book Acoustic Phonetics on page 303. Then, in 2005, Bachrach, Lulich, and Malyska reported on a speech perception study in which they manipulated the frequency of the subglottal zero (the pole-zero pair was simplified to an effective zero) in synthetic vowel stimuli. They found that the frequency of the subglottal zero influenced whether the vowel was perceived as a front or a back vowel.

Updated: 12/16/05