Intelligent Subgrouping of Multitrack Audio

Subgrouping facilitates the simultaneous manipulation of a number of audio tracks and is a central aspect of mix engineering. However, the decision process of subgrouping is a poorly documented technique. This research sheds light on this ubiquitous but poorly de ned mix practice, provides rules and constraints on how it should be approached as well as demonstrates its bene t to an automatic mixing system. I rst explored the relationship that subgrouping has with perceived mix quality by examining a number of mix projects. This was in order to decipher the actual process of creating subgroups and to see if any of the decisions made were intrinsically linked to mix quality. I found mix quality to be related to the number of subgroups and type of subgroup processing used. This subsequently led me to interviewing distinguished professionals in the audio engineering eld, with the intention of gaining a deeper understanding of the process. The outcome of these interviews and the previous analyses of mix projects allowed me to propose rules that could be used for real life mixing and automatic mixing. Some of the rules I established were used to research and develop a method for the automatic creation of subgroups using machine learning techniques. I also investigated the relationship between music production quality and human emotion. This was to see if music production quality had an emotional e ect on a particular type of listener. The results showed that the emotional impact of mixing only really mattered to those with critical listening skills. This result is important for automatic mixing systems in general, as it would imply that quality only really matters to a minority of people. I concluded my research on subgrouping by conducting an experiment to see if subgrouping would bene t the perceived clarity and quality of a mix. The results of a subjective listening test showed this to be true