Speaker Recognition Based On Vector Quantization

Speaker recognition is an important part in the processing of automatical recognizing included in speech signal,It has broad application foreground in many fields,It can automatically judge if or who is the speaker in the people by analyzing the characteristic parameters.One of the most important questions in the speaker recognition is the selection,extraction of feature parameters.The features we choose should can represent the speaker's voice characters and be insusceptible of environments, robust,keeping acceptable performance for different users and also can be used under the normal background noise level. In this paper,we use full pole model to obtain speech signal characteristic. Before picking up the speech signal characteristic parameters ,the voice signal is undergoing pretreatment, in order to analyse speech characteristic parameters and compare with the basis methods of speaker recognition,we choose mel-frequency ceptral coeffients and linear prediction ceptral coffients's difference to be the speech characteristic parameters.we use Vector Quantization to recognize and utilize Matlab Voice Box to abstract speech's characteristic parameter.we improve traditional method and prove that it is better than traditional one.