The difficulty of detecting errors in one's own reports may be the bottleneck in enabling a widespread application of automatic speech recognition (ASR). Although quite a lot of research concerning correction procedures has been conducted, detection strategies have received little attention. As ASR systems still produce errors a lot depends on the detectability of errors especially in the medical environment, as reports have to be error-free. This paper presents the results of an experiment which investigated error detection performance during a diagnostic task. No differences in detection performance could be found between auditory or visual feedback except for the detection of substituted words. For this category of errors, visual feedback proved to be better.
|Title of host publication||Proceedings of the 4th European Conference on Speech Communication and Technology, (Eurospeech-1995), Madrid, Spain, 18-21 September 1995|
|Publication status||Published - 1995|