Embedded Voice Recognition System Research

Along with rapid development about computer software and hardware technology, semiconductor technology, electronic technology, communication technology and network technology etc, human being community is going to the behind epoch about PC. Recently years in the field of biological recognition technology, the voice recognition technology is being concerned by the advantage of especially convenience, economical, accurate, and become the most important and universal security authentication model in human being's daily life and work. However plenty of real-time data collection, filtration, disposed demand new processors that have high-float calculate speed, low-power to achieve.Today digit signal processor owing to its specially instruction structure and high compile efficiency can execute the voice recognition arithmetic quickly, can satisfy various digital signal process and high real-time demands at present, especially for the field of vector calculation, point linearity addressing etc, it has a overwhelming advantage. Because of the voice recognition technology's convenience, economically, accurately and embedded system's portable, mobility, the voice recognition system that using the high quality digital signal processor is well known by human beings, and is widely being used in people's daily life, become the main trend of development of science and technology in the future. The main research direction of this thesis is to realize speaker identification function under the embedded platform, the main works of this thesis are as follows:(1) On the TMS320C6713 develop starter kit research board realize voice signal collection function, utilize serial port to store the digital voice signal on time; by the ensuring of higher rate of correct identification, utilize appropriate voice print recognition algorithm, realize voice signal practice and recognize function.(2) Combination the voice signal's collection module, practice module, recognize module into a suit of integrated voice recognition system, having high rate of correct recognition. By utilizing cross compile tool, according to software develop kit that is supplied in different embedded platform rebuild software module, it insure the system having high transplant ability.