Male/female speech classification based on cepstral modulation ratio parameterization by Laguerre polynomials

This paper uses a new set of feature vectors that is based on modulation spectrum of cepstral coefficients by means of Laguerre regression method. The performance of the proposed method is investigated by a gender classification of a noisy speech. Compared with other regression methods, our proposed feature set demonstrates high performance in the sense of gender classification. Low classification errors obtained in different noisy scenarios proves the superiority of the new feature vectors for the classification task.