Research On Speaker Recognition Algorithm

Speaker recognition (SR) technique is one of biological technologies with which computer can automatically identify the individual based on human characteristics. Compared with traditional methods, the new method proposed in this paper is more convenient, safer, and not easy to be forgotten or replaced. Speaker recognition can be used under a lot of circumstances, such as the telephone electronic commerce, military wiretapping, information retrieval, and so on. Speaker recognition technology has been made on certain research results. There are some major difficulties that confront large extractive feature data, which will consumes large memory and long computing time to all speech parameters. This makes Real-time implementation very hard and expensive. Thus the problem of improving train time and recognition time has turned into the most active research filed without deteriorating recognition performance.In this paper, we have stdueid emphatically that how to improve train time and recognition time in order to meet Real-time implementation of speaker recognition without deteriorating recognition performance, and discussed related problems to speaker recognition. My work of this paper is outlined as the following:Firstly, in the speaker recognition, there are some major difficulties that confront large extractive feature data, which will consumes large memory and long computing time to training SVM with all speech parameters. Speaker recognition based on improved PSO-SVM approach is proposed, a particle swarm optimization (PSO) method is proposed with adaptive inertia weight by the change of the number of iterations based on the analysis of inertia weight global best fitness of the PSO. The improved PSO increases the ability to avoid local optimum. Then a speaker recognition method using this improved algorithm to train support vector machine is presented. The experimental results show that the presented SVM method optimized by PSO for speaker recognition can achieve higher recognition accuracy and higher recognition speed.Secondly, To increase the recognition speed of speaker identification, a novel feature extraction method based on the kernel K-means clustering and the sparse kernel principal component analysis (SKPCA) for speaker recognition was proposed: Here kernel K-means clustering is to divide all the frames of each sample into a given amount of clusters, since the resulted clustering centers can represent better the clusters they belong to, the clustering is replaced by the clustering center, and the dimensions of kernel matrix are decreased accordingly. This method reduced storage and computational complexity, meet the requirement of speaker recognition in terms of practicability.