Equal-loudness contours measured by the randomized maximum likelihood sequential procedure

When the equal-loudness level is measured by the method of constant stimuli (CS), the point of subjective equality (PSE) can be strongly influenced by the setting of sound pressure levels of the variable stimuli. This effect is called the range effect. To eliminate the range effect, the maximum likelihood sequential procedure in which a randomization process is introduced may be effective. We named this method the randomized maximum likelihood sequential procedure (RMLSP). To study the robustness of RMLSP against the range effect, psychophysical experiments were carried out. The results showed that the range effect had a strong influence on PSEs obtained by CS, while its effect on those obtained by RMLSP was negligible. A set of equal-loudness levels was also measured using RMLSP. The results at frequencies below 1 kHz were significantly higher than those specified in ISO 226-1987. This is a consistent tendency which has been observed in all recent studies. From the results of comparisons between PSEs obtained in our previous studies using CS and those of the present study, it was shown that the bias of the range effect did not show up in most of the "averaged" PSEs measured in our previous studies. Moreover, a part of the previous data that significantly differ from those obtained by RMLSP could be corrected based on a simple model of the range effect.