Science & Engineering Research Support soCiety (SERSC)
Several speaker identification systems are giving good performance with clean speech but are affected by the degradations introduced by noisy audio conditions. To deal with this problem, the authors investigate the use of complementary information at different levels for computing a combined match score for the unknown speaker. In this paper, they observe the effect of two supervised machine learning approaches including Support Vectors Machines (SVMs) and Naive Bayes (NB). They define two feature vector sets based on Mel Frequency Cepstral Coefficients (MFCCs) and RelAtive Spectral TrAnsform Perceptual Linear Prediction coefficients (RASTA-PLP).