Disorder Speech Classification for Clinical Data Using SVM

In this work, mild level of mental retardation (MR) children speech samples were taken for consideration. In the existing system, there are many effective treatments for the problem of stammering. Most of these involve making changes in the manner of speaking. They are conducted by speech and language pathologists by giving fluency in speech practice in general. The proposed work is, the acute spot must be identified for affording speech training to the speech disordered children. In this paper, still classification of speech is found. Initially Feature Extraction is implemented using Mel Frequency Cepstrum Coefficients (MFCC) for both words of normal and pathological subjects' speech. Dimensionality reduction of features extracted is implemented using Principal Component Analysis (PCA). Finally the features are trained using Top of Form Support Vector Machines (SVM) for classification.