Comparació entre diferents sistemes d’extracció de característiques per la classificació d’Instruments (MIR)

Abstract

[en] In this Final Degree Project, we conducted an experiment comparing two sound feature extraction methods, the first of which is based on the Mel scale (MFCC), and the second on Wavelet Transforms (WTCC). The experiment will determine which system is better at categorizing musical instruments, a job that may be extremely useful in areas such as audio post-production, recommendation algorithms, music analysis, and so on. This study is a continuation of Angel Bergantiños and Chan Yoon's TFG, in which algorithms were constructed to classify musical songs based on their musical genre using supervised and unsupervised machine learning approaches, respectively.