Development of a revised loudness model.

Abstract

The loudness model developed in my laboratory was based on models developed by pioneers such as Fletcher and Zwicker. Their models involved calculation of an excitation pattern for the sound of interest, from the long-term average spectrum of the sound, transformation of the excitation pattern to a loudness pattern (specific loudness plotted as a function of frequency on a perceptually based frequency scale), and calculation of loudness from the area under the loudness pattern. The innovations used in the Cambridge model were: (1) Incorporation of a linear filter as a first stage, to account for the transmission of sound through the outer and middle ear; (2) Calculation of excitation patterns from extensive measurements of auditory filter shapes, using the notched-noise method; (3) Transformation of the frequency scale to the ERBN-number scale, which is similar conceptually to the Bark scale, but is different numerically, especially at low frequencies; (4) Transformation of excitation to specific loudness using a function such that loudness has a finite value at the absolute threshold, as opposed to zero loudness. More recently, the model has been modified to include the concept of binaural inhibition, to accommodate experimental evidence indicating that loudness does not simply sum across ears.