Title :
Improved Linear Prediction Technique for Musical Onset Detection
Author :
Lee, Wan-Chi ; Kuo, C. -C Jay
Author_Institution :
University of Southern California, Los Angeles, USA
Abstract :
Following our previous work in [8] where a music signal is decomposed into multiple subband signals and the forward linear prediction errors of all subbands are combined to give the detection function to localize the onset position, two enhanced features are examined in this work. First, we propose a new method to decide the stepsize, which is a critical parameter in LPEF adaptation. Second, we consider a more flexible and robust subband decomposition scheme that is adaptive to the characteristics of the music signal. It will be shown by experimental results that these modifications enhance the accuracy of the onset detection scheme.
Keywords :
Adaptive filters; Adaptive signal detection; Bandwidth; Multiple signal classification; Narrowband; Nonlinear filters; Predictive models; Robustness; Signal processing; Signal processing algorithms;
Conference_Titel :
Intelligent Information Hiding and Multimedia Signal Processing, 2006. IIH-MSP '06. International Conference on
Conference_Location :
Pasadena, CA, USA
Print_ISBN :
0-7695-2745-0
DOI :
10.1109/IIH-MSP.2006.265058