Recall of passages of synthetic speech

Abstract

Memory for synthetic speech versions of grade school-level materials was tested in two studies. In Experiment 1, two different versions of three simple stories were recorded in synthetic speech. The versions differed in prosody, one employing hand-applied pitch and stress and the other employing random stress. Free recall of the stories showed no consistent difference in performance as a function of the intonational pattern used. Experiment 2 compared the recall of a simple biographical sketch presented in either natural speech or synthetic speech. A sentence-by-sentence dictation test showed little difference in intelligibility of the texts, and the difference disappeared with minimal practice on synthetic speech. Free recall of the entire passage showed synthetic speech to be disadvantaged only in the case of nonpracticed listeners. Again, minimal practice with synthetic speech dispelled the differences.