Voice Pathology Detection Using Decision Tree Classifier
Fahad Taha AL‐Dhief, N. M. Abdul Latiff, Nik Noordini Nik Abd Malik, Marina Mat Baki, Naseer Sabri, Musatafa Abbas Abbood Albadr, Nurul Fariesya Suhaila Md Sazihan
Abstract
Nowadays, the systems of voice disorder detection obtained considerable attention due to the high importance of this field. However, the assessment of voice pathology requires certain tools and well-trained doctors. Moreover, this assessment can be identified by a group of professionals who listen to a patient in order to assess the patient's speech to identify whether the patient's voice is pathological or normal. Nevertheless, this assessment is based on the listener’s experience. Consequently, machine learning is the most suitable technique for the detection of voice pathology, where this technique is a cost-effective and non-invasive method. Therefore, this paper presents Decision Tree (DT) algorithm based on the Mel-Frequency Cepstral Coefficient (MFCC) technique for the detection of voice pathology. The voice samples for pathological and healthy classes are collected and taken from the Saarbrucken Voice Database (SVD). The performance of the DT algorithm is assessed in terms of many evaluation measurements such as accuracy, sensitivity, precision, G-mean, F-measure, and specificity.