| Nowdays, information networking and economic globalization is changing continuously. The traditional identity authentication methods bring inconvenience, easily lost, forged, and passwords cracked easily, such as ID card, key, smart card, passwords, etc. These methods are low security and poor reliability, but biometric-based authentication has advantages of unique, invariant, imitability easily. Biometric identification technology is key to authentication field in many countries. In the area of biology characteristic recognition technology, the speaker recognition is attached more importance by common people because of its unique convenience, efficiency and accuracy superiority.In this paper, electronic-sound lock is one application of speaker recognition technique. When the system starts to recognize, firstly processes the voice signal, then matches with characteristic vector in the template storehouse, finally exports the recognition result to decide weather to lock. After a study of speech signal processing, the paper analyzes the method and the process of character extraction and does lots of research on vector quantization methods as well. Later, a vector quantization based authentication algorithm is presented and fellowed by its simulation in MATLAB, the result of which shows that the algorithm works well in speaker recognition. In hardware design aspect, the paper gives the corresponding hardware design solution after knowing the characteristic of the TMS320VC5502chip, including circuit design of each block in details. In the software realization aspect, the process of the corresponding software is discussed according to speaker recognition algorithm as well as corresponding hardware designs based on DSP/BIOS operating system.By debugging and analyzing the experimental result, the correctness and feasibility of the vector quantization based on authentication algorithm, and the accuracy of system design, and the stability of hardware platform and software operation is verified. |
PDF Full Text Request