Musical Style Identification with n-Grams and Neural Networks
Pedro P. Cruz-Alcázar, María José Castro-Bleda · Lecture notes in computer science · 2008
Musical Style Identification (MSI) aims to automatically classify music by style. It is being recently explored, mostly in the field of multimedia databases, with potential applications to content-based retrieval. But MSI may be also employed in other applications. We try to face up this challenge with two different methodologies: n -gram Models and Neural Networks. Very good results were obtained with n -grams in our previous research and we were willing to test how other Artificial Intelligence techniques performed with this task, so we began a preliminary study with Multilayer Perceptrons that is promising.