Accent Recognition of Speech Signal Using MFCC-SVM and k-NN Technique

feature reduction, feature categorization, and feature selection were among them for calculating efficiency. The SVM and k-NN classification algorithms, along with the MFCC feature extraction methods, are presented in this article for accent-dependent speaker recognition systems. The classification task is achieved using two training and testing stages. In the training stage, the parameter of speaker-specific features is calculated and different speaker statistical model is generated. In the testing stage, the unknown speaker's speech sample is compared with the speaker's statistical model and then classified using different classifiers. If the amount of accent is dependent on language, then it is becoming more crucial accent recognition. In the Indian language southern part of India is extensively spoken. It has dissimilar accents. The spoken languages of Rayalaseema, Telangana, and coastal Andhra are the most prominent accents. A sample of speech is collected from resident speakers of the Telugu language in three distinct accents as part of the technique that has been proposed. It is used for both training and testing features.