Perceptual learning of synthetic words and sentences
Steven L. Greenspan, Howard C. Nusbaum, David B. Pisoni · The Journal of the Acoustical Society of America · 1985
Previous research has shown that the intelligibility of synthetic speech can be improved with training [E. C. Schwab, H. C. Nusbaum, and D. B. Pisoni, Human Factors (in press)]. In the present study, we investigated the relationship between the type of training subjects receive and the pattern of perceptual learning that occurs. Three groups of subjects were given word recognition tests consisting of synthetic words and sentences generated by the Cottax Type-'n-Talk, before and after a training period. One group of subjects was trained on isolated synthetic words. A second group was trained on fluent synthetic sentences. A control group received no training at all. Recognition of isolated synthetic words improved equally for both groups trained on synthetic speech compared to the control group. However, overall word recognition in sentences only improved for subjects that were trained on synthetic sentences. These results demonstrate that performance with synthetic sentences will predict performance with isolated words, but the converse may not always hold. [Work supported by AFOSR and NIH.]