A computational model of loudness density

Loudness density is the distribution of loudness across dimensions of time and frequency. A computational model of loudness density is described that reproduces equal-loudness level contours from time-domain representations of tones. The model also demonstrates increase in loudness as the bandwidth of a multi-tone complex increases, an observation that is consistent with psychophysical measurements. The model provides an efficient method for computing the loudness of arbitrary acoustic waveforms. When loudness is assumed to be the decision variable for psychophysical tasks such as intensity discrimination and masked thresholds, model-predicted performance can be compared with psychophysical measurements. The similarity between model predictions and psychophysical estimates of suppression is noteworthy because the model provides no separate mechanism for suppression. Suppression in the model is only possible through the same mechanism that provides compression for single tones. Its ability to predict suppressive effects suggests that it also may be a useful tool for exploring more general mechanisms of auditory signal processing.