K-Means Clustering and Support Vector Machine for Assamese Dialect Identification

Assamese, spoken by the inhabitants of Assam in the northeastern region of India, belongs to the larger Indo-European language family. The Indo-European language family includes the Assamese language. Assamese dialects are created when Assamese speakers in various regions of the state use different vocabularies, syntaxes, and pronunciations. We provide our findings in this study on recognizing spoken data in four Assamese dialects: Eastern dialect, Central dialect, Kamrupia dialect, and Goalporia dialect. The automatic dialect identification system described in this study extracts features using K-means clustering on Mel-Frequency Cepstral Coefficients (MFCCs) and classifies them using a Support Vector Machine (SVM). The complexity of the SVM classifier is greatly decreased when the MFCC features are post-processed using K-means clustering before being sent to the classifier. This is otherwise impossible to prevent because each voice signal contains a substantial number of MFCC features. The efficiency of the suggested strategy is assessed using a database of recorded speech in four dialects of Assamese: Eastern, Central, Kamrupia, and Goalporia. The recommended approach obtains an average classification accuracy of 87% using short-duration voice inputs.