ABSTRACT
The critical band as it relates to simultaneous tone-in-noise masking is reevaluated in light of the results from two experiments. The first measures the detectability of a 500 Hz tone masked by white Gaussian noise when the overall level of the stimulus is randomized from interval to interval and trial to trial. The results indicate that subjects can use information that is outside the critical band centered at the signal frequency, or outside the temporal interval that contains the signal, to overcome the performance decrement caused by randomizing level. The second experiment investigates the detectability of a 500 Hz tone masked by reproducible noise. A multiple channel model, fit to the data, suggests that subjects compare information in different spectral/temporal regions of the stimulus in order to determine the presence of the signal.
