Rap-Style Commentary Generation in an Audience Participation Fighting Game

Abstract

This paper presents a rap-style commentary system for enhancing audience participation games. In this work, an audience participation fighting game played by two AI players is presented. This game has two unique features. The first is that commentaries are generated in a rap-style fashion. The second is AI-player strength adjustment based on chat messages from audiences-these chat messages also affect the abovementioned commentaries. Experimental results show that the proposed system significantly promotes audience experience.