Nyquist plot and quantitative characterizations of acoustic voice signals for speaker recognition

The “Nyquist plot” is a visual analytical method we previously developed for identifying various pathological glottal conditions when applied to high-speed digital imaging (HSDI) derived glottal signals and acoustic signals. In this study, we applied the Nyquist plot to acoustic voice output aiming to: 1) identify differences between the properties of five vowel sounds commonly used in human language, and 2) characterize the voice signals of individual subjects. We also explored numerical parameters derived from spectral analysis of the signals to be used synergistically with the Nyquist plot. We found that the Nyquist plot displays distinct patterns for each of the five vowel phonations investigated, and that plots also displayed potential for the identification of individuals when used in combination with additional numerical parameters. However, with a sufficiently large data set it would likely be necessary to use multiple phonations in order to uniquely identify an individual. Expected potential for development are in the fields of computer-based recognition and understanding of human language, clinical diagnostics, and security and identification of human individuals based on the unique properties of their voices.