The System For Access Control Based On Speaker Recognition And Its DSP Implementation

Speaker recognition is the important content of speech processing. It is widely used in many fields such as access control, telephone bank, security, military and national defence. The algorithm of speaker recognition is studied and its DSP implementation.In algorithm aspect, vector quantization with small template and small quantity of computing is used in this thesis. Algorithm is simulated in Matlab and gets the good result. C language programming for the algorithm is implementation in TI's Code Composer Studio.In DSP implementation aspect, the system for access control based on digital signal processing chip TMS320DM642 is designed. The whole system included two parts: speech acquisition and processing part and control and display part. Speech acquisition and processing is finish by DSP board. In this thesis, DSP'serial port communicating, designing and managing template library, reading/writing/erasing FLASH, designing the Boot loader program, format converting from .out to .hex, burning FLASH are implemented. A control-display board is designed for displaying the recognition results and controlling the whole system. This board includes nine keys, a LCD and a serial port. Three keys are defined that is recognition keys, training keys, template eraser keys. The other keys are reserved. This board communicates with DSP board by serial port. The whole system can run independence.In this thesis, system running independence is implemented. It is practical in access control based on speaker recognition and it contributes a lot for apply the speaker recognition in real life.