Cross-validation of electroglottographic based analysis using glottal area waveforms

We present our latest findings on clustering-based electroglottographic (EGG) and glottal area waveform (GAW) analysis of vocal fold vibratory states in professional tenors. The EGG signal, representing the time-varying contact area as the vocal folds vibrate gives information only about the closed or contact phase of vocal fold vibration. Conversely, the GAW representing the time-varying area between the open vocal folds, can provide information only for the open phase. The great advantage of the EGG is that it is very easy to acquire, in contrast to the GAW, which requires invasive procedures. To assess the validity of EGG-based analysis, we compare the accuracy of vibratory state differentiation using clustering with EGG-based, GAW-based and combined EGG & GAW-based features. The small difference between the error rates of the three different analysis settings, indicates that EGG-based analysis is in most cases sufficient for the differentiation of vibratory states.