Speech-based alarm displays

Dr Chris Baber, et al. have shown that performance using synthesised speech is inferior to that using human speech, especially if the tasks involve processing of the message rather than simply spotting words in the message. The time interval between an alarm appearing and a participant manually accepting the alarm were logged by the computer. The mean time to accept an alarm in the speech alarm condition was 11.68 seconds. The main difference in performance occurs at the ‘accept’ stage, resulting from a delay between the alarm and action in the speech condition. If the speech display places an attentional demand on the operator, dealing with the alarm display itself becomes a problem. This would result in not only a time delay pertaining to the length of the speech message, but also an additional time delay resulting from processing and handling the message. Text-based displays present alarm messages on a visual display unit's page within a list of other messages.