Automatic background music selection for tabletop games

• Ano de defesa: 2018
• Autor(a) principal: Padovani, Rafael Rodrigues
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Viçosa
• Palavras-chave em Português: Processamento de linguagem natural (Computação); Aprendizado supervisionado (Aprendizado de máquina); Jogos para computador - Canções e musica; Música; Ciência da Computação
• Link de acesso: http://www.locus.ufv.br/handle/123456789/23909

System accuracy is a crucial factor influencing user experience in intelligent inter- active systems. Although accuracy is known to be important, little is known about the role of the system’s error distribution in user experience. In this dissertation we show, in the context of background music selection for tabletop games, that su- pervised learning algorithms can make the system “indecisive” by performing errors that are sparsely distributed in a game session. We then introduce Bardo, a real- time intelligent system to automatically select the background music for tabletop role-playing games. Bardo selects and plays as background music a song represent- ing the classified emotion. With variants of Bardo we also introduce an ensemble approach with a restrictive voting rule that instead of erring sparsely through time, it errs consistently for a period of time. We show that our ensemble approach is able to make the system decisive. We hypothesize that sparsely distributed errors can harm the users’ experience and it is preferable to use a model that is somewhat inac- curate but decisive, than a model that is accurate but often indecisive. A user study in which people evaluated edited versions of the D&D videos suggests that Bardo’s selections can be better than those used in the original videos of the campaign. A second user study was performed following the same process and the results suggest that understanding how different error distributions affect user experience is key to develop intelligent systems able to successfully interact with humans.