The long-term objective of this research is to develop a scientific basis for standardization of acoustic voice analysis. This type of analysis is being used today to test various aspects of human body function and condition. Attempts are being made to study mental disorders, neurological disease, aging, emotion, veracity, stress, and a host of respiratroy and larygeal disorders. Effects of smoking, alcohol, drugs, and pollution on the human voice have also received increased attention. There are no agreed-upon standards, however, for eliciting and recording voice samples, or for extracting acoustic measures. Furthermore, there is no theoretical framework for interpretation of the measures. This reserach contributes toward the development of such standards and interpretations.
Specific aims are to develop a set of vocal utterances that test the range and stability of the vocal fold oscillator when muscle tensions, airway structure, and viscoelastic properties of mucus and tissue are abnormal. A series of steady phonation, phonatory transitions, and sentences are proposed within the framework of the phonetogram. Redundancies are eliminated between perturbation measures and spectral slope measures that parametrize the acoustic signal. Since some of the acoustic effets are subtle, criteria are examined for adequate fidelity in recording, filtering, and sampling of voiced utterances. Questions regarding length and multiplicity (number of tokens) of acoustic observations are raised and answered. Finally, a data bank of natural and synthesized vocal utterances in initialized for eventual distribution. As it expands, it will serve as a benchmark for calibration of procedures, for training of clinicians, and for preceptual studies on the human voice.
Showing the most recent 10 out of 11 publications
