Confidence Ratings and Message Reception for Filtered Speech

The statistical decision model, which has achieved outstanding success in describing the detection of signals in noise, was applied to the reception of filtered speech and the relation between correct confirmations and false alarms obtained with filtered speech corresponds with that typically obtained with noise interference.

Abstract

The statistical decision model, which has achieved outstanding success in describing the detection of signals in noise was applied to the reception of filtered speech. A confidence rating was added to the articulation test procedure in order to obtain additional information about the listener's criterion of message acceptance and message rejection of filtered speech. The relation between correct confirmations and false alarms—the Receiver Operating Characteristic—obtained with filtered speech corresponds with that typically obtained with noise interference. It is suggested that the “noise” of the decision model may be extended to a wide range of operations which perturb the signal.