Two classifiers score fusion for text independent speaker verification

For text-independent speaker verification, the Gaussian mixture model (GMM) using a universal background model (UBM) Strategy and the Support Vector Machines (SVM) are the two most commonly used methods. Recent approaches dealing with speaker and channel variability have proposed the idea of stacking the means of the GMM model to form a mean super vector. The new model introduces the resulting super vector to SVM system. The main contribution of this paper is the investigation of the performance gained using data fusion strategy. Indeed, we applied three fusion methods to the GMM-UBM and the GMM-SVM systems. Experimental results, on speaker database, show that a significant improvement is observed compared to baseline method.