Laryngeal Evaluation Using Features from Speech and the Electroglottograph

Signal processing techniques were utilized to discriminate speakers with laryngeal pathology from a normal population. Classical periodicity analyses and discriminant analyses were performed separately on a set of features extracted from two data sets taken from both subject populations. The two data sets consisted of the acoustic and electroglottographic signals recorded simultaneously during an experimental paradigm. Preliminary results encourage further refinement of the technique and use of the electroglottograph in this application.