PNCC attributes for robust recognition of independent speaker from the text

Abstract

Automatic speaker recognition has been a subject of intense research over the past decade. However, its performance under state-of-art algorithms is drastically degraded in presence of noise. This paper focuses on the application of the novel technique Power-Normalized Cepstral Coeficients (PNCC) to the text-independent speaker recognition task. Our aim of this study is to evaluate the performance of this feature in comparison with the conventional Mel Frequency Cepstral Coeficients (MFCC) and Gammatone Frequency Cepstral Coeficients (GFCC) methods.