Robust glottal activity detection using the phase of an electroglottographic signal

Abstract

Abstract Electroglottography (EGG) is a non-invasive technique of studying the glottal activity. The phase of EGG signal has not attracted much attention, although it is robust and capable of revealing dynamics of glottal activity. The main objective of this work is to develop a robust and accurate technique for analyzing the glottal activity by exploiting the phase of EGG signal. Using the phase of EGG signal, accurate and robust methods are proposed to detect glottal closure instant (GCI) and glottal opening instant (GOI) within a glottal cycle. The glottal activity during close and open phases of a glottal cycle can be precisely analysed by using GCIs and GOIs respectively. Robustness of the proposed method for detection of GCIs and GOIs is also evaluated in vulnerable cases such as transition in laryngeal mechanism, weak voicing and voice transition. The accuracy of the proposed method is evaluated using manually labeled EGG signals chosen from CMU Arctic database, and the performance of the proposed method is also compared with the state-of-the-art singularity in EGG by multiscale analysis (SIGMA) method. Finally, the performance of the proposed and SIGMA methods is also evaluated on EGG signals recorded from the patients having vocal folds disorders.