Study on Closed-set Speaker Identification Basedon Biomimetic Pattern Recognition

In this paper, a new classifier of speakeridentification has been proposed, which is based onBiomimetic pattern recognition (BPR). Distinguishedfrom traditional speaker recognition methods, such asDWT, HMM, GMM, SVM and so on, the proposed classifier is constructed by some finite sub-space which is reasonable covering of the points in high dimensional spaceaccording to distributing characteristic of speech featurepoints. It has been used in the system of speaker identification. Experiment results show that better effect couldbe obtained especially with lesser samples. Furthermore,the proposed classifier employs a much simpler modelingstructure as compared to the GMM. In addition, the basicidea "cognition" of Biomimetic pattern recognition (BPR)results in no requirement of retraining the old system forenrolling new speakers.