A Novel Algorithm for Detecting Spasmodic Dysphonia Voice Pathology using Random Forest Frame Work

Spasmodic dysphonia, a rare voice disorder is detected in the current work using Random Forest frame work. Voice pathology is related to the vocal tract area affecting the quality of speech. Numerous voice pathologies have been over the years of them are unnoticed as the symptoms are not significant. Even the symptoms are known the nature of the disorder is difficult to identify due to the over lapping nature of the symptoms. The existing algorithms for voice pathology detection are capable of classifying between normal and affected subjects, while the nature of the disorder has been considered in the proposed algorithm. Computational complexity has been reduced due to the incorporation of finite significant energy features estimated over non overlapping frames. Classification of accuracy of 93.5 has been seen with a population of 100 trees.