Thuryalankara: artificial intelligence based audio plugin for Sri Lankan percussion instruments

Abstract

Sri Lankan music is yet to prove its musical prowess by incorporating artificial intelligence tools, therefore, this research introduces a novel invention, an automated audio plugin for music producers, so the process of creating, mixing, mastering, and producing music is easier. To achieve this, the research introduces a Variational AutoEncoder (VAE) machine learning model to create and generate music, an artificial intelligence (AI) system that can automate the mastering process. This research also introduces an innovative component, a virtual instrumentation tool using MIDI technology for the Sri Lankan percussion instruments that allow users to play the instrument virtually using a MIDI keyboard, and alongside it, a preset beat generator that automatically maintain tempo consistency. Thuryalankara was able to receive a collective average of 80% accuracy rate exceeding the predicted accuracy rate of 65% from the software benchmarking test and the physical survey conducted with music producers. Finally, with the inclusion of powerful tools like this, the ultimate objective of this research is to take the Sri Lankan instruments to the international level where any producer from little to plenty experience is able to use this plugin to enhance their musical production.