A Frequency-Weighting Digital Filter in Sound Level Meter Based on Neural Computing Method

Abstract

Frequency weighting networks are a critical component of a sound level meter (SLM), and their error characteristics directly determine the performances of SLM. For reducing the high-frequency error of the [Formula: see text] frequency-weighting filters with the bilinear transformation method (BTM), a design method for [Formula: see text] frequency-weighting filters based on neural computing method (NCM) is proposed. A detailed algorithm for solving the filter coefficients is provided, and the amplitude-frequency characteristics of the [Formula: see text] frequency-weighting filters with BTM and NCM are compared in detail. The experimental results show that the amplitude-frequency characteristics of the [Formula: see text] frequency-weighting filters in SLM with NCM are significantly better than those of BTM. The filter meets the requirements of the first class SLM defined by IEC61672, which demonstrates the effectiveness of this proposed method.