基于分布式算法的声像定位

Localization of Sound Image Based on Distributed Arithmetic

根据声像定位应用及FPGA内部结构的特点,对分布式算法的实现做了若干优化,设计了一个具有通用性的高性能声像定位协处理器,并用FPGA最终实现了对声像的虚拟定位.经Xilinx公司FPGAxc5vlx110t验证,即使采用多达512点的HRTF,其处理能力依然可达264.061 Mbit/s,比未经优化的实现节省资源72%.虚拟定位的声像经真人试听测试均包含正确的位置信息,且对原声的失真较小.

A general and high-performance coprocessor, which implements localization of sound image based on FPGA, is designed by considering the characters of localization of sound image and architecture of FPGA, in order to optimize the implementation of distributed arithmetic. Verified by FPGA xc5vlx110t of Xilinx, coprocessor can,even dealing with up to 512 points HRTF, produce 264. 061 Mbit/s data rate with 72% inside resource cost reduced comparing to the implementation before optimizing. Virtual localized sound images have also been approved to contain correct localization information and very few distortion of the source by testing on real people.