Automatic cymbal classification using non-negative matrix factorization

Several musical instrument classifiers have been proposed. While many approaches in sound-feature extraction and in sound classification have been successfully used, most focus on distinguishing different harmonic instruments such as the violin and the flute, whose sounds have very different characteristics. On the other hand, much less attention has been given to percussion instruments, especially if we consider the discrimination of instruments of the same type, like the cymbals in a drum kit. Here, we propose a classifier that is able to distinguish this latter type of instruments. The classifier is able to distinguish sounds with very similar properties, like sounds produced by instruments with similar geometry that differ in material or size. In particular it is able to distinguish sounds from the cymbals in a drum kit. Instead of using a set of predefined features, the classifier learns spectral features from the data using non-negative matrix factorization. This work is important to fill the gap on percussion instrument classification and transcription (since most music transcribers focus on harmonic instruments).