1 Department of Architecture, Design and Media Technology, The Faculty of Engineering and Science (ENG), Aalborg University, VBN2 Sektion København, The Faculty of Engineering and Science (ENG), Aalborg University, VBN3 The Faculty of Engineering and Science (TECH), Aalborg University, VBN4 University of Sussex5 University of Sussex
Dictionary-based methods (DBMs) provide rich possibilities for new sound transformations; as the analysis dual to granular synthesis, audio signals are decomposed into `atoms', allowing interesting manipulations. We present various approaches to audio signal cross-synthesis and cross-analysis via atomic decomposition using scale-time-frequency dictionaries. DBMs naturally provide high-level descriptions of a signal and its content, which can allow for greater control over what is modified and how. Through these models, we can make one signal decomposition influence that of another to create cross-synthesized sounds. We present several examples of these techniques both theoretically and practically, and present on-going and further work.
Proceedings of the International Computer Music Conference, 2011