Machine Learning Approach for Assamese Dialect Identification
摘要
The goal of this work is the identification of spoken data in four Assamese dialects: Eastern, Central, Kamrupia, and Goalporia. The research aims to enhance technology accessibility for speakers of local dialects while also contributing to the preservation of these unique linguistic variations. To achieve this, automatic dialect identification systems were employed, which analyze voice signals to ascertain the specific dialect being spoken. An audio file and its speaker can be identified by determining the voice signal's frequency and loudness, among other factors. In this research, Mel frequency Cepstral coefficients (MFCC) was utilized to capture relevant aspects of speech signals for dialect identification. Accuracy rates of 78% and 75%, respectively, were obtained by applying support vector machines and decision tree classifiers as classification approaches.