A system for masking audio signals includes a microphone for generating an ambient audio signal representing ambient noise, a speaker for rendering masking audio, and a processor in communication with the microphone and the speaker. The processor performs spectral analysis on the ambient audio signal from the microphone to determine a spectral envelope of the ambient noise, adjusts a frequency response of an optimizing filter based on the spectral envelope, applies the optimizing filter to a baseline masking waveform, producing an output waveform with relative spectral distribution matching the ambient noise, and provides the output waveform to the speaker.