Using non-speech sounds to increase web image accessibility for screen-reader users

Abstract

Screen-reader users access images on the Web using alternative text delivered via synthetic speech. However, research shows that this is a tedious and unsatisfying experience for blind users, because text-to-speech applications lack expressiveness. This paper, poses an alternative approach using an experiment that compares audemes, a type of non-speech sounds, with alternative text delivered using synthetic speech. In a pilot study with fourteen sighted users, findings show that audemes perform better across many areas. Specifically, audemes required lower mental and temporal demands and led to less effort and frustration and better task performance. Moreover, participants recognized audemes with higher accuracy and lower errors. Audemes were also perceived as more engaging compared to alternative text delivered using synthetic speech. Additionally, audemes were found to be richer in delivering information. This study suggests that non-speech sounds could substitute or complement alternative text when describing images on the Web.