MoSievius: Interactive Audio Mosaicing

This thesis investigates how audio analysis techniques and representations can be used to automate the sequencing of arbitrary sounds in order to provide a higher level and more expressive control over the creation of sample based music. This pursuit required the creation of a mosaicing algorithm prototyping language, which expedites the process of creating and testing different ways to interactively control the creation of audio mosaics. A number of different mosaicing algorithms were written and tested within this framework that gave insight into the efficacy of different audio representations as both a control mechanism for creating audio mosaics and as a mapping between raw samples and their cumulative perceptual effect.