Automatic speech recognition in the medical environment

Abstract

Medical specialists are interested in applying Automatic Speech Recognition to save time and money spent on medical reporting. A number of issues need to be resolved in order to apply this technology successfully. This paper concentrates on the issue of feedback. An experiment is described in which the most appropriate feedback modality for presenting the recognition result in the situation of the pathologist is investigated. Although at this moment visual feedback seems the safest solution, error-detection performance is still poor. A totally different approach to medical reporting is presented which may prove to be the best solution.