Design and Evaluation of a Spoken-Feedback Keyboard

Abstract

Speech recognition technologies have come a long way in the past generation. Indeed, they are becoming ever more pervasive in our day-to-day lives, especially in the form of voice-activated menus so prevalent in many automated answering systems. However, speech technologies are still of limi ted usefulness for large-vocabulary speaker-independent applications in noisy environments, especially where relatively limited computing resources are available as in present-day personal digital assistants (PDAs). Given the popularity of digital cellular phones and text-messaging systems, we describe a generic interface that can be used by any application that need text input by visually-impaired users on this kind of devices. Given the shortcomings of present-day speech recognition technology, we opted to develop three types ofkeyboards, two predictive, with vocalfeedback. This paper, describes the interface development and the usability evalualion results with target users. Our prototype testing scenario included composing short text messages (SMS), and sending them via digital cellular networks (GSM) making it accessible to visually-impaired people.