An Improved Feature Extraction Method in Speaker Identification

With the rapid development of information and computer technology, speaker recognition technology has obtained more and more research and application. In this paper, we put forward a dynamic fuzzy clustering algorithm based on principal components analysis- PCA-DFC. In order to reduce the characteristic dimension and computational complexity, principle component analysis method was used to process front speech signal. Through dynamic fuzzy clustering, different categories can be obtained from different λ∈[0, 1], so that is the dynamic clustering. Then F-statistics Method was applied to choose the optimal classification threshold λ. Experimental results show that less characteristics parameters are employed in this method, it can obtain higher identification rate and well reduce the feature dimension compared with the traditional method.