Design And Implementation Of Speaker Recognition System On FPGA

Speaker Recognition indentify or verify the identity of the speaker based on the personality characteristics included in the speaker’s voice. Speaker Recognition has received extensive attention because of its unique advantages, after years of research and exploration,there have been many mature Speaker Recognition system.FPGA can achieve high speed processing because of its parallel processing structure, so it can meet the market’s request of the real-time on the system. In addition, FPGA using SRAM technology, can be unlimited programming, which is helpful to system upgrade and maintenance. The Speaker Recognition system of the paper uses FPGA as the hardware platform.First, on the basis of further rsearch of Speaker Recognization theory,this paper design a Speaker Recognization system suitable for FPGA implementation, and simulate and verify the designed Speaker Recognition system based on VQ using Matlab,then complete each FPGA function module of Speaker Recognizon by a variety of logic design technologies, in the same, verify its function with the help of simulation tool of Modelsim.The main content is as follows:(1) Analyze the purpose and significance of the subject,discuss the development and research status of Speaker Recognition, and introduce the main voice feature parameters and Speaker Recognition algorithm.(2) According to FPGA structure, select the Speaker Recognition based on VQ which is easily realized on FPGA.Introduce each composition module of Speaker Recognition system. According to the traditional VQ Speaker Recognition system, add the process of voice enhancement, and modify the distortion measurement as weighting distance distortion measurement. Using Matlab,generate the Speaker Recognition model through training and verify system’s speaker recognition effect.(3) Analyse the basic principle of FPGA,elaborate Xilinx’s FPGA development kits ISE Design Suite10.1,and explain simply the hardware platform of Speaker Recognition,XUP Virtex-II Pro development board, which makes a ready for the FPGA implementation of Speaker Recognition system. (4) Based on the top_to_down module design rule,design each logic sub_module of Speaker Recognition based on FPGA, which include Pre_sharp module, enframe and window module, endpoint check module, voice enhancement module, VQ module and judgement module. Accomplish functional simulation and logical validation of the all modules using simulation tool of Modelsim.(5) Design control module, which can make each system functional sub_module work harmonily in order to achieve Spearker Recognition, and display the recognition result,which is transferred by RS232serial interface from FPGA to PC,using super terminal.Make FPGA functional simulation and verification on the whole system using Modelsim, and analyse the simulation result.