摘要
MDCT和IMDCT凭借其在消除边界噪声、改善信号质量方面的优越性,广泛应用于高级音频编码标准框架中,同时,其也是编解码系统中计算复杂度最高的三大模块之一。在Clenshaw′s递归公式的基础上,采用关键单元分时复用、信号并行处理技术对共用递归中滤波处理模块进行改进,在总资源几乎不增加的条件下,使得总递归次数降至N2/16+N/4,且模块化规则性好、结构性强;在硬件实现时,增加延时单元和适当控制逻辑以避免并行数据处理带来的多端口或多个存储单元的并行操作。最后,使用SynplifyPro8.5.1工具在XilinxVirtex-E系列FPGA上对36点共用结构进行了验证。
The MDCT （Modified Discrete Cosine Transform）/IMDCT is popular due to its advantage in reducing bound noise and improving quality of audio signal. A solution based on Clenshaw＇s recurrence formula is proposed, in which the circuit-sharing and parallel processing technology are used to reduce the total executed numbers. Its regularity and modularity make it suitable for VLSI implementation. In addition, two delay-cells and some control logic are used for replacing a multiport memory. Finally, experiment on Xilinx FPGA indicates that the calculation on first-half and second-half index time-domain data can be realized.
出处
《电声技术》
2009年第3期68-70,73,共4页
Audio Engineering
