Language Impairment Detection Based on Speech Signals and Gaussian Naive Bayes algorithm

Speech impairment has been seen in children more frequently now a days due to various reasons. It is necessary to identify speech impairment early to provide necessary treatment for early development. In this work, a method has been proposed for speech impairment detection using tonnetz feature of speech and Gaussian naive Bayes (GNB) algorithm. The speech signals are processed and the tonnetz features are extracted from the speech signals. The tonnetz factors are then used with the GNB algorithm to detect speech impaired or not impaired. The result of the method shows an average accuracy of 92.38%. Hence, the proposed speech impairment detection method can be used for monitoring the speech of the children.