 |
    |
Automatic Han Chinese Folk Song Classification Using Extreme Learning Machines
Suisin Khoo, Zhihong Man, and Zhenwei Cao
Faculty of Engineering and Industrial Sciences, Swinburne University of Technology, VIC 3122, Australiasuisin.khoo@gmail.com
zman@swin.edu.au
zcao@swin.edu.au
Abstract. Multilayer feedforward neural networks trained via supervised learning have proven to be successful in pattern recognition. This paper presents the technique of using single hidden layer feedforward neural network as an automatic classifier in music classification. Han Chinese folk songs from five distinct geographical regions in China are studied and encoded using a novel musical feature density map (MFDMap) for machine classification. The extreme learning machine (ELM) and its two variants are employed as the classifiers to categorize the folk songs. Our simulations show that by using a low-pass finite impulse response extreme learning machine (FIR-ELM), we can achieve 80.65% classification accuracy.
Keywords: Musical feature density map, music classification, extreme learning machine, artificial neural network, Han Chinese folk song
LNAI 7691, p. 49 ff.
lncs@springer.com
© Springer-Verlag Berlin Heidelberg 2012