A sound detecting mechanism capable of forming a diaphragm and a back electrode on a substrate by a simple process includes acoustic holes corresponding to perforations formed on a front surface of a substrate. A second protective film, a sacrificial layer and a metal film are laminated on the front surface in a portion corresponding to the acoustic holes. The substrate is etched from the back surface thereof to a depth reaching the acoustic holes to form an acoustic opening. Subsequently, by effecting an etching from the back surface of the substrate through the acoustic holes, the sacrificial layer is removed and a void area is formed between the diaphragm made of the metal film, the substrate and the formed perforations. The sacrificial layer that remains after the etching is used as a spacer for maintaining a gap between the back electrode and the diaphragm.