Northwestern Univ., Evanston, IL 00208
A set of tests are described in which the effects of echo-free audio delay on PCM-encoded speech are evaluated. In these tests, a conversation-type subjective speech assessment technique is used, which is augmented with several objective measures of conversational performance. At the same time, a measure of mental workload is studied to assess the cognitive effects of audio delay on the human end user. The workload measure that is used is a secondary task involving short term, or working, memory. In describing these experiments, relevant background information is provided on audio delay, which illustrates the difficulties in measuring user acceptability of audio delay. Information on human memory models and key experimental findings are reviewed, which motivate the choice of workload measure used in this study. The test methodology and results are then reviewed in detail. User tolerance of audio delay is found to decrease with increasing audio delay, in accordance with other researchers findings. In addition, added cognitive load is demonstrated in the presence of long delays via decreased working memory performance.