In this article, we propose an audio equalizer of musical genres\nbased on fuzzy logic. Widely used audio playback software,\nsuch as VLC or iTunes, propose genre-specific equalization\ncurves to be applied for the whole duration of the music.\nThese curves are the same for all songs belonging to a\nspecific genre, and they do not take into account the specifics\nand diversity of each song. We propose a different strategy.\nResearch in music information retrieval has revealed a significant\nnumber of audio descriptors that allow for the recognition\nand description of diverse musical genres. We use some\nof these descriptors to feed a fuzzy logic inference system,\nwhose outputs are the required equalization levels for each\nfrequency band. The rules of the system are derived from the\nanalysis of a well known music database encompassing ten\ndifferent musical genres. Our results indicate that our approach\nworks for songs that exhibit multiple genre characteristics,\nthat are difficult to classify into one category, or that\nmix genres.
Loading....