|Title||Short-term feature space and Music Genre Classiﬁcation |
|Publication Type||Journal Article |
|Year of Publication||2011 |
|Authors||Marques, G., Langlois T., Gouyon F., Lopes M., & Sordo M. |
|Journal Title||Journal of New Music Research |
|Abstract||In music genre classiﬁcation, most approaches rely on statistical characteristics of low-level features computed on short audio frames. In these methods, it is implicitly considered that frames carry equally relevant information loads and that either individual frames, or distributions thereof, somehow capture the speciﬁcities of each genre. In this paper we study the representation space deﬁned by short-term audio features with respect to class boundaries, and compare diﬀerent processing techniques to partition this space. These partitions are evaluated in terms of accuracy on two genre classiﬁcation tasks, with several
types of classiﬁers. Experiments show that a randomized and unsupervised partition of the space, used in conjunction with a Markov Model classiﬁer lead to accuracies comparable to the state of the art. We also show that unsupervised partitions of the space tend to create less hubs.
|Full Document||http://www.tandfonline.com/doi/pdf/10.1080/09298215.2011.573563 |