Evaluation of Feature Learning Methods for Voice Disorder Detection

Voice disorder is a frequently encountered health issue. Many people, however, either cannot afford to visit a professional doctor or neglect to take good care of their voice. In order to give a patient a preliminary diagnosis without using professional medical devices, previous research has shown that the detection of voice disorders can be carried out by utilizing machine learning and acoustic features extracted from voice recordings. Considering the increasing popularity of deep learning, feature learning and transfer learning, this study explores the possibilities of using these methods to assign voice recordings into one of two classes—Normal and Pathological. While the results show the general viability of deep learning and feature learning for the automatic recognition of voice disorders, they also lead to discussions on how to choose a pre-trained model when using transfer learning for this task. Furthermore, the results demonstrate the shortcomings of the existing datasets for voice disorder detection such as insufficient dataset size and lack of generality.