Wavelet analysis of turbulence in the glottal jet flow

Abstract

Airflow from the lungs drives the vocal folds into oscillation and exits the glottis as a turbulent jet, even though laminar flow is documented entering the glottis. Incomplete glottal closure may result in air leakage that becomes turbulent. This turbulence is thought to be the main source of breathiness in the voice. Few studies have measured degree of turbulent noise in the glottis. This turbulent noise adds significant amounts of energy to the voice signal and cannot be extracted easily without changing the voice. In this study, the turbulent jet was investigated in an excised canine larynx model with simultaneous recordings of air particle velocity, subglottal pressure, airflow rate, and EGG signal for various conditions of phonation. The velocity was measured with a constant-temperature hot-wire anemometer system. To separate the turbulence from the periodic component of the velocity signal, the technique of wavelet denoising was employed. In this method, unlike the Fourier transform which decomposes the signal into periodic components, wavelet denoising uses template matching transform to approximate the signal with different resolution or scale. Results indicate that the high resolution part contains the turbulent noise. [Work supported by NIDCD Grant No. R01 DC03566.]