The Role of Audio Features in Accent Recognition: A Comparative Analysis

This study focuses on enhancing Automatic Speech Recognition (ASR) systems, crucial in Science and Technology, by addressing challenges tied to diverse speaker accents. The research delves into accent recognition, comparing Mel Frequency Cepstral Coefficients (MFCC) and Chroma as feature sets. Surprisingly, Chroma surpasses MFCC in accent classification. This study widens our understanding of accent recognition techniques, aiding ASR accuracy. By investigating MFCC and Chroma, finding an optimal feature for superior accent recognition, thus enhancing ASR system performance with varied accents.