Recognized Based On The Identity Of The Voiceprint Identification System

Voiceprint recognition technology is also called the speaker recognition technology with the figurative language. It is the technology to determine the identity of the speaker. It bases on the characteristic parameters of human psychological and physiological. These parameters are reflected in the waveform generated by the people when they speak. Voiceprint and fingerprint are exclusive. Because people own different natural features. So voiceprint and fingerprint are unique. Voiceprint recognition technology can be used as an important means of identification. Because the voice contains the individual's natural attribute and bright individual color. In addition, the acquisition of speech signals is very convenient. The price of the collection equipment is not expensive generally. These advantages are superior to identify the identity of the speaker by the analysis of voice characteristics. As a security authentication technology, speaker recognition with its unique convenience, economy and accuracy will bring people a new value.Although the study is largely consistent. Speaker recognition and speech recognition are still different. It mainly concerns the natural characteristics of the people not the speaker's words.The study focuses on the speaker's personality traits. Speaker recognition is divided into speaker verification and speaker identification.This paper is the text-dependent speaker verification system. After the description of the acoustic basics of speech signal processing and speech signal generation, front-end processing,endpoint detection for the voiceprint identification system is made. It also introduces speech endpoint detection method with dual-threshold two sentences and the application. It compares LPC parameters and MFCC parameters. MFCC cepstrum parameters are used to calculate the desired feature parameters. Finally, a variety of recognition model are introduced. HMM model is the most typical application. It is one of the widest forms of recognition models. This paper focuses on the HMM model. HMM model is used to achive a digital speech recognition simulation in MATLAB. The recognition rate of the experiment system reaches 90%, which verifies the recognition accuracy of HMM model.