Classification of Healthy and Pathological Voices Using Artificial Neural Networks

Speech is the basis of communication between people. In daily life, function loss occurs in mechanisms that create voice due to reasons such as occupation, environment, age and gender. In this study, 57 healthy and 150 pathological voice data (from Hyperkinetic Dysphonia, Hypokinetic Dysphonia, Reflux Laryngitis) was classified using proposed fetaures and Artificial Neural Networks (ANN). The data obtained from voice data is given as input to ANN model. Depending on these input data, the output information of the artificial neural network is determined as patient or healthy. In order to classify the patient and healthy group, two hidden layers and an output layer were used as the artificial neural network model with the least error. As a result of the study, the accuracy of classification between patients and healthy groups was 90.47 %.