Extended Average Magnitude Difference
Function Based Pitch Detection
Ghulam Muhammad
Department of Computer Engineering, King Saud University, Saudi Arabia
Department of Computer Engineering, King Saud University, Saudi Arabia
Abstract: This paper presents a new extended average magnitude difference function for noise robust pitch detection. Average magnitude difference function based algorithms are suitable for real time operations, but suffer from incorrect pitch detection in noisy conditions. The proposed new extended average magnitude difference function involves in sufficient number of averaging for all lag values compared to the original average magnitude difference function, and thereby eliminates the falling tendency of the average magnitude difference function without emphasizing pitch harmonics at higher lags, which is a severe limitation of other existing improvements of the average magnitude difference function. A noise robust post processing that explores the contribution of each frequency channel is also presented. Experimental results on Keele pitch database in different noise level, both with white and color noise, shows the superiority of the proposed extended average magnitude difference function based pitch detection method over other methods based on average magnitude difference function.
Keywords: Pitch detection, AMDF, EAMDF, and noise robust.
Received May 13, 2009; accepted January 3, 2010