Nonlinear Source-Filter Interaction in Singing
Ingo R. Titze, Ph.D.
Distinguished Professor of Speech and Voice, University of Iowa
Executive Director, National Center for Voice and Speech
The classical linear source-filter theory of voice production is compared to a nonlinear theory that includes interaction between the vocal tract and glottal airflow. By including the subglottal tract, it is shown that the combined vocal tract impedance can be primarily inertive (exhibiting positive reactance) over a wide range of Fo. The epilarynx tube and the glottal entry tube contribute to this heightened inertive reactance and serve as impedance matchers for the rest of the vocal tract. It is shown that a high degree of glottal flow waveform skewing is attributable to source-tract interaction, with skewing quotients ranging from 1.0 to 10.0. To a large degree, epilarynx tube diameter governs waveform skewing and the peak glottal flow. For high levels of interaction, a speaker or singer can conserve glottal flow without loss of maximum flow declination rate (MFDR). Thus, the interactive system is more efficient from a flow consumption point of view, but it does not yield favorable conditions slightly above the first formant frequency, where the combined subglottal and supraglottal reactance is compliant for a small range of fundamental frequencies. Inferences are drawn with respect to voice quality adjustments for speaking and different styles of singing.