Main Article Content
This paper compares machine learning algorithms in how they classify audio signals into musical genres. A literature review of previously present techniques and approaches is also done, based on feature engineering and algorithms. Unique dataset is formed using an online GTZAN music database, with sampling and other techniques to balance classes. The audio analysis library is used for extracting key features from the samples.
Analyses are then performed, for each feature and genre. Classifiers are created based on machine learning concepts (SVM and NN) and training and testing are performed on algorithms, with the dataset made. In the end, a wholistic process evaluation is conducted (an assessment of the features and dataset chosen and other parameters of the project) to infer an outcome.
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.