Human-Centred Artificial Intelligence in Concatenative Sound Synthesis

Abstract

Concatenative Sound Synthesis (CSS) is a data-driven method to synthesize new sounds from a large corpus of small sound snippets. It unlocks endless possibilities of re-creating sounds, which is exciting, particularly as the technique also necessitates very little musical knowledge for anyone to utilize. However, synthesizing a specific sound does require more than just matching segments of sound at random. Very few synthesis results are the result of the first raw output. Sometimes, sound manipulation and transformation need to be applied to the synthesized sound. At other times, the desired results can only be achieved by adding or removing certain sound segments in the corpus collection.