声纹认证前端音频预处理SOPC设计

Design of an audio processing filter based on SOPC for voiceprint extracting front

针对声纹身份认证一体化设计需求,基于SystemVue和Quartus Ⅱ平台,利用DSP Builder技术,在EP2C35F芯片上完成了用于声纹特征提取的前端音频预处理模块的SOPC设计。主要通过SystemVue设计数字音频滤波器参数,然后结合Simulink平台完成基于线性相位结构和分布式(DA)算法的低通滤波器结构设计,最后通过Signal Compiler转换为HDL语言完成设计。实验结果表明,系统可以有效限制音频信号的带宽,减少带外噪声;利用DSP Builder设计具有线性相位结构和DA算法的滤波器,设计周期短,占用逻辑资源少,其运算速度有很大提高;MATLAB理论计算也验证了设计的正确性。

For voiceprint authentication integrated design requirements, front audio preprocessing module for feature extraction of voiceprint SOPC designs was implemented in EP2C35F using SystemVue, Quartus Ⅱ platform and DSP Builder technology. Digital audio filter parameters were designed primarily through SystemVue. And then with Simulink platform ,the low-pass filter was de- signed based on linear phase structure and distributed （DA） algorithm. Finally, the design for the HDL language was given through Signal Compiler conversion. The results showed that the system can effectively limit the bandwidth of the audio signal to reduce noise band, and using DSP Builder to design the filter with linear phase structure and DA algorithm maked sure design cycle was short, occupied logic resources were few, and the operation speed was greatly improved. The correctness of this design was verified by MATLAB theoretical calculations also.