一款可编程语音处理器的设计与应用

Design and Application of Programmable Speech Processor

为了提高通信系统的保密性，降低制造成本，需要进行专用处理器的设计。该文基于SELP(Sinusoidal Excitation Linear Prediction)算法模型原理，设计了一款高质量多速率语音专用处理器芯片。芯片使用可重构体系结构和超长指令字系统设计方法，将复杂度高的子程序进行优化，能够显著提高指令并行度。仿真结果表明：在该芯片上实现语音压缩编码算法，执行效率高于相同工艺水平的通用数字信号处理器，并保持原有编码质量。该处理器能够实现多种类型的语音压缩算法，使语音算法可以达到高保密性、低复杂度和易开发性。
关键词：语音信号处理；语音压缩；专用芯片；可重构体系结构；超长指令字
【Abstract】Programmable application-specific processor is necessary in order to increase communication system security and reduce manufacturecost. An application special processor is introduced for high quality multi-rate speech compression algorithm based on SELP(sinusoidal excitationlinear prediction) model. The chip uses reconfigurable architecture and VLIW(very long instruction word) system design method, by which highcomplexity sub programs are optimized to get a remarkable instructions parallel. The simulation result shows speech compression algorithms whichare implemented on this chip have higher efficiency than that on universal DSP(digitalsignal processor) under the same technical level, whilemaintaining the original coding quality. The chip can implement different types of speech compression algorithms and reach to good security, lowcomplexity and development convenience.
【Key words】Speech signal processing; Speech compression; Application-specific integrated circuit(ASIC); Reconfigurable architecture; Very longinstruction word

可重构体系体系结构，是指某一计算系统能够利用可重用的硬件资源，根据不同的应用需求，灵活地改变自身体系结构，以便提供与之相匹配的体系结构[1]。已有可重构体系实现了语音算法的某些特定部分，如自适应多速率算法中的代数码本搜索[2]、原型编码器中的小波变换[3]以及有限脉冲响应滤波器[4]等，但使用协处理器使硬件规模仍然偏大，且只能对一类语音算法适用。如能在整体芯片设计中采用可重构体系，能够大幅度降低开发时的工作量，同时避免引入协处理器所导致的规模大、保密性差等缺点。本文基于这条技术路线，设计了一种可编程专用语音编解码芯片TR100。