非线性量化的模拟实现方法

Study of non-uniform quantization analog-method

基于语音信号的A律或μ律非线性量化方法,由于其先采用一个线性转换器进行量化,再用数字硬件或软件量化为A律或μ律电平。其为:用较少的二进制位代替较多二进制位的数字压缩编码。在动态范围更大的非语音信号采集中,常被误认为可以用此法来"扩展小信号的可量化范围",因前线性转换器的量化台阶决定了其最小信号分辨率,这里的问题主要在:信号的量化前后处理意义不同。要扩展小信号的量化范围,在A/D转换的位数一定情况下,为保证输入峰值不超载,应该对其输入的模拟信号进行A律或μ律折线非线性放大处理。

A method of A-law or μ-law-based non-linear speech signal quantization is presented,which employs a linear converter for quantization first,and then uses digital hardware or software quantization to generate A-law or μ-law signals.i.e.a digital compression encoding scheme uses fewer bits representing more bits.In data acquisition of speech signals with a wide dynamic range,it is often believed mistakenly that this scheme could be applied to＂expand the quantifiable range forsmall signals,＂due to the fact that the minimal resolution of small signals is determined by the quantization steps of the front-end converter,the problem mainly lies in the difference in meaning between the front-end quantization of signals and the quantization afterwards.To expand the quantifiable range of small signals,with a given number of A/D conversion bits,the input analogue signals should be amplified with A-law or μ-law non-linear processing,in order to ensure that the peak of input signals are not over-loaded.