DocumentCode
1712417
Title
Melodic pitch extraction from music signals using modified group delay functions
Author
Rajan, Rajeev ; Murthy, Hema A.
Author_Institution
Department of Computer Science and Engineering, Indian Institute of Technology Madras, Chennai India
fYear
2013
Firstpage
1
Lastpage
5
Abstract
In this paper, we propose a method to extract the melody pitch in a polyphonic environment, using phase information of the signal. Group delay functions are used for melody extraction. The group delay function computed directly from the speech signal has been widely used to extract source and system parameters. The main objective of this study is to address melodic pitch extraction using features based on the Fourier Transform phase of a signal rather than the conventional Fourier Transform magnitude. As the phase is difficult to process, the modified group delay function is processed to extract pitch. The performance of the proposed algorithm is evaluated on ADC-2004 dataset, LabROSA training set and MIR-1K dataset. The results of the proposed method are comparable to that of existing methods based on the Fourier transform magnitude spectrum.
Keywords
Cepstral analysis; Delays; Feature extraction; Fourier transforms; Multiple signal classification; Music; Speech; Cepstral smoothing; Group delay; Modified group delay;
fLanguage
English
Publisher
ieee
Conference_Titel
Communications (NCC), 2013 National Conference on
Conference_Location
New Delhi, India
Print_ISBN
978-1-4673-5950-4
Electronic_ISBN
978-1-4673-5951-1
Type
conf
DOI
10.1109/NCC.2013.6487986
Filename
6487986
Link To Document