Computer, Run Program : Assessing Viability of Audio Interfaces for Interactive Fiction

Abstract

Interactive fiction is a genre with a lot of potential and very enthusiastic communities. In this thesis, I investigated technological innovations in the genre, and then made an attempt on an experimental system of a fully voice-based interactive fiction game that is both controlled by speech and gives output by speech, without requiring a visual user interface. This did not wholly succeed, since the system provides some basic command-line output, but both the speech control and the speech output are functional, albeit not completely without problems. For the purpose of building the system, I looked into different products that do the required functions for the features: speech recognition, speech generation, and narrative engines for interactive fiction. Then I tested players’ experiences with the system, by recruiting volunteers, having them play the game and reporting their experiences with it. The results remained inconclusive due to the small number of testers, but the data showed that further research might be in order, since none of the testers rejected the format outright. The results also gave ideas on how to improve the system further. The voice output is not ideal, and the speech recognition for the speech control could be more accurate. While the narrative engine serves its purpose for simple games, it is not suitable for more complex stories. For more conclusive research in the future, these problems should be addressed.