Modeling microphone in PSpice based on Neural Network

Presently, the electret condenser microphone (ECM) is used in almost every consumer and communication audio application. In order to improve the efficiency of the circuit design, circuit simulation is necessary. In this article, we present a method based on Neural Network for modeling ECM, by which the fundamental characteristics of an ECM, including sensitivity, directivity, output impedance and frequency response, are modeled in PSpice. Firstly, each input-output characteristic is approximated with different Neural Network, after which the structures, weights and biases of the Neural Network depicting different characteristic of the ECM are acquired. Secondly, the structures are described in PSpice language to form sub-circuits respectively. Finally, these subcircuits are integrated into a unitary sub-circuit based on the relationship between the inputs and output of the ECM to form the final model.