The Application of Fusion Technology for Speaker Recognition

This paper describes the fusion technology for speaker recognition. First, these feature sets are extracted form the same speech signal, and then the correlations between the feature sets are analyzed, only the uncorrelation feature sets can be combined. The single feature set and combined feature set are compared. Experiments show that the performance of combined feature set is better than the single feature set. In the model fusion technology, different speaker recognition models sets are combined. The performance of the combined model is compared to other Models. Results show that the combined model can improve the performance of speaker recognition system.