Research Of Speaker Recognition Based On Platform Of Embedded System

Biometrics technology has a broad application foreground in many fields such as electric business and information security etc. As a kind of biometrics technology, speaker recognition is to identify the speaker's identity by distilling the speaker's information from phonic signals. Compared with the other biometrics technologies, speaker recognition is more convenient, more accurate and more economical.First, the valid acoustic features and methods of speaker recognition are introduced, and the common embedded-system hardware platforms are compared. Then, the concept and structure of the embedded-system hardware platform are presented, and the platform's operation system, processor and debug interface circuit of the system are introduced.LPCC is adopted as characteristic parameters. Support Vector Machines (SVM) is a classifier devised for nonlinear and high dimensional sample problems, and has an effective classify and learning ability, so speaker recognition models based on SVM are designed. LPCMCC is utilized as another characteristic parameter, which can reflect the hearing frequency nonlinear characteristic.The ARM9 and Linux are chosen to make up the embedded-system hardware platform. Boot Loader and Linux kernel are designed, the file system is built up, and the SVM is transplanted successfully into the embedded-system.