ABSTRACT
This chapter discusses the overall goal of how a computer program can simulate human-like sound design processes. An example using music information retrieval methodologies demonstrates developing a computational model for machine listening of aesthetic criteria for a Creative A.I. system. Creative tasks, such as graphic design, video production, game design and music-making, have grown to be the leading multi-modal professional use of computers. Recent initiatives in artificial intelligence and machine learning are driving a renaissance in how creative tasks are carried out by amateurs and professionals alike. For example, it is now possible to generate hundreds of hours of music or playable game levels in only minutes. However, generating the sound design for entertainment media remains an open problem. The challenges provided by computational sound design rest upon the subjective experience of the listener response, the continuous signals of the media and the complexity in the alignment of sound elements with the richness of the narrative and visual environment. Given the importance of sound design in the user experience and the growing amount and sophistication in production, it is critical to advance the research and development of computationally assistive technologies in this field.
