Comparing virtual patients with synthesized and natural speech

Abstract

Virtual Patient (VP) simulations are often designed to use pre-recorded speech in order to provide more realism and immersion. However, using actors for recording these utterances has certain downsides. It can add to the cost during implementation, can take considerable time especially when a large number of VPs have to be created, and is not very flexible for example when sentences or words have to be added frequently. This study aims to explore the use of synthesized speech as an alternative to pre-recorded speech for VPs. Two medical scenarios have been prepared for this study, and both have been implemented using a VP with natural language or with synthesized speech. In a pilot study we explored students' retention rates of the symptoms reported by the VP under both conditions to investigate whether synthesized speech can serve as a good enough alternative.