An electrostatic speaker system including at least one electrostatic speaker element having a pair of stators and a diaphragm therebetween, each of the stators and the diaphragm having an least one electrically conductive portion. The conductive portion of the diaphragm is patterned as a mesh, which may include gaps, and a conforming layer overlying the conducting portion of the diaphragm and disposed so as to cover gaps in the mesh. A speaker drive circuit provides soft clipping of the audio input signal so that the audio signal applied between the diaphragm and the stators does not exceed the maximum acceptable level that can be applied between the diaphragm and the stators of at least one speaker element.