On the use of NMF for onset detection in poliphonic piano music

Author

Costantini, Giovanni ; Todisco, Massimiliano ; Saggio, Giovanni ; Perfetti, Renzo

Author_Institution

Dept. of Electron. Eng., Univ. of Rome "Tor Vergata", Rome, Italy

fYear

2011

fDate

19-21 Sept. 2011

Firstpage

1

Lastpage

4

Abstract

Recently the use of non-negative matrix factorization (NMF) for music onset detection has been proposed. In this paper we present the results of an experimental evaluation of NMF for onset detection of piano music. The test results, concerning a publicly available dataset of piano music, show a good performance of NMF, almost invariant from the factorization rank. However, better results have been obtained using simplified onset detection techniques, with lower computational complexity of implementation.

Keywords

computational complexity; data analysis; matrix decomposition; music; computational complexity; dataset; factorization rank; nonnegative matrix factorization; poliphonic piano music; simplified onset detection techniques; Computational complexity; Detection algorithms; Discrete Fourier transforms; Educational institutions; Machine learning; Signal processing; Simulation; constant Q transform; non-negative matrix factorization; onset detection;

fLanguage

English

Publisher

ieee

Conference_Titel

Intelligent Signal Processing (WISP), 2011 IEEE 7th International Symposium on

Conference_Location

Floriana

Print_ISBN

978-1-4577-1403-0

Type

conf

DOI

10.1109/WISP.2011.6051698

Filename

6051698