Development and evaluation of an adaptive loudness scaling procedure

Abstract

Loudness scaling procedures are used in the diagnostics of hearing impairment to assess the recruitment phenomenon, i.e., the pathological growth of loudness. In these procedures, acoustical stimuli at different levels are presented to the subject and the task is to rate the perceived loudness on a category scale. The aim of this study is the development and evaluation of an adaptive loudness scaling procedure in which the level of the next stimulus is calculated on the basis of the previous trials. Based on empirical results from normal-hearing and hearing-impaired listeners, the procedure was optimized in order to yield minimal test–retest deviations and minimal biases due to range and context effects at minimum measuring time. To achieve these aims, different settings of the following parameters were tested: Number of response alternatives, track length, initial level, adaptive procedure, and parametrization of the loudness function. The best setting was shown to yield an equal distribution of stimulus levels in the full dynamic range of the individual subject with a minimum number of ‘‘inaudible’’ and ‘‘too loud’’ presentations. For each categorial loudness the corresponding stimulus level is estimated with a standard deviation of about 5 dB.