Automatic source identification of monophonic musical instrument sounds

Author

Kaminsky, I. ; Materka, A.

Author_Institution

Dept. of Electr. & Comput. Syst. Eng., Monash Univ., Clayton, Vic., Australia

Volume

1

fYear

1995

fDate

Nov/Dec 1995

Firstpage

189

Abstract

A system has been developed which automatically identifies the source of monophonic musical instrument sounds. Preprocessing of sound recordings includes calculation of the short term RMS energy envelope, principal component analysis and ratio/product transformations of the resultant principal components. An artificial neural network and a nearest neighbour classifier were compared to determine which one provided optimum classification ability. The system performance was tested on sounds recorded from four musical instruments chosen to represent each of the major musical instrument families and playing notes over the range of one octave under varying volume conditions. Classification accuracies in the range 93.8-100% were achieved

Keywords

acoustic signal processing; backpropagation; multilayer perceptrons; music; musical instruments; pattern classification; RMS energy envelope; automatic sound source identification; backpropagation; monophonic musical instrument sounds; multilayer perceptron; nearest neighbour classifier; principal component analysis; ratio/product transformations; sound recordings; Acoustic testing; Acoustical engineering; Artificial neural networks; Australia; Instruments; Low pass filters; Power engineering and energy; Principal component analysis; System performance; System testing; Systems engineering and theory;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks, 1995. Proceedings., IEEE International Conference on

Conference_Location

Perth, WA

Print_ISBN

0-7803-2768-3

Type

conf

DOI

10.1109/ICNN.1995.488091

Filename

488091