The Research On The Method Of Speaker Recognition And The Realization Of DSP

The work of this paper focuses on the method of speaker recognition,mainly researching on the vector quantization algorithm using MFCC features and the decision in speaker verification . On the basis of the study mentioned above, it also researches on the open-set speaker recognition method from the definition of the set, and presents an open-set speaker recognition algorithm based on DTSV and realizes with Digital Signal Processor (DSP). Besides, design and accomplish a speaker Real-time recognition system, which can recognition a speaker in the real-time. This system has been used in a phonetic electronic lock in the floor door elementarily.Recently in the field of Biometric Identification Technology, Voiceprint Recognition, which is often said speaker recognition, draws much of people's attention, with its facility, economy and precision, which causes it to become an important and popular safe verification way in our daily life and work increasingly.There is much information in the speech signal. Speaker recognition is an area,which recognizes speaker by extracting speaker's information automatically from his speech signal. Speaker recognition is a branch of speech signal processing, it involves a lot of fields, such as the acoustics, phonetics, linguistics, physiology, psychology, information engineering, communication theory, electronic technique, computer science, pattern recognition, artificial intelligence etc, and is a interdisciplinary application field. It has developed quickly within this decade and has important application feature in military, law and bank, etc.In the information field, digitalizing becomes a mainsteam in modern information processing. The digital signal process (DSP) technique, one of the core techniques in digitalizing, has become most advanced branches, which people is paid much attention to and develops so rapidly. At the same time, it has become a new technical field and an independent academic subject system, which makes its application scope and market much more widely. The DSP technique takes on the digitalizing industry revolution task, therefore, DSP is the key of this revolution. It can be seen that the research on the speaker recognition and the realization of DSP having important theory and application value. The main work and content are sum-up as follow:The dissertation consists of six chapters. In Chapter 1, it explains the dissertation research contents and principal work. At the same time, we briefly review the development and present situation information of Biometric Identification Technology , speaker recognition and DSP.The basic theory study is described in Chapter 2. It introduces the basic knowledge of speech signal processing and the elemental theory t of speaker recognition. At the same time, it summarizes the basic of the software realization of DSP .All the above contents are the base of later research.In Chapter 3, researching the method of speaker recognition, it mainly includes the method using MFCC as speaker's features. The speaker recognition method based on vector quantization is simple to calculate, without thinking about the problem of complicated statistical model and the time wraping. At the same time, this method can make a better performance, when the training data we get are less. So the method of vector quantization is widely used in the field of speaker recognition. Mel Frequency Cepstrum Coefficient describe non-linear character of human ears perception to frequency by using Critical- Band effect of hearing system. At last ,carry out the algorithm simulation, and get a good performance. In Chapter 4, research on the method of decision in speaker verification, mainly including the method of making a decision based on the information of candidate sequences of speech segment identification, the range of the distortions and speaker verification with dynamic thresholds and do simulation of these algorithms. It also introduces a method of on-line incremental threshold update, which provides a promising way for...