Creating the Benchmark Dataset for Ethno-Fusion Music: From Piano Rolls to an Ensemble of Instruments

There's much to explore in the space where music meets computer science, with numerous challenges when integrating music with computer technologies. This paper introduces the development of a specialized dataset for music genres that features distinctive melodies, emphasizing the critical use of the MIDI format for capturing notes and frequencies essential for computational and mathematical models to produce accurate outcomes. We've crafted a dataset encompassing 1000 unique melodies across various rhythms like Arabic, reggaeton, and the 7/8, time signature, employing MIDI piano-roll to ensure the precision of melody representation. These melodies were compared with performances on other instruments, which were primarily utilized for solo pieces. Aimed at aiding computer programs in accurately utilizing musical frequencies for a range of purposes, this dataset emerges as a pivotal resource. Moreover, the melodic lines created are unparalleled, marking the inception of the first corpus in this vein crafted with such high precision. It concludes that until now, there has not been a collection of MIDI melodic lines of this format, underscoring the novelty and uniqueness of our dataset.