摘要
介绍了作为语音识别系统的基本环节之一——采样前置带通滤波器的研制情况,对采样前置滤波器的滤波特性的确定原则进行了讨论,提出应如何选择带通滤波器的上限截上频率和采样速率,确保混叠发生在感兴趣频带之外,借以减小混叠误差。所研制的采样前置GIC 椭圆带通滤波器由八阶高通和九阶低通滤波器串联构成,所用的运算放大器数目较多。为了得到预期的滤波特性,保证各运算放大器工作在合理的状态,作者设计了零电平调整电路,使得滤波器的调试变得容易,性能指标满足要求。
A new-type elliptical GIC bandpass filter, as an important component forspeech recognition, was developed to precede sampling. The new filter is conposedof a 8-order high pass and a 9-order low pass in series connection, with a nu-mber of operational amplifiers used. To attain the expected filtering characte-ristics.it is necessary to ensure all the operational amplifiers in well reasonableworking condition. For this purpose, a zero--level adjusting circuit was desig-ried to enable the filter to be debugged easily with performance targets satisf-actorily reached.
关键词
采样前置
带通滤波器
impedance converter
aliasing and aliasing error
common mode voltage input
zero level adjusting circuit