Improved variable ZA-LMS algorithm based on discrete cosine transform for high noise reduction

In the strong noisy environment of oil well equipment, normal voice communication shall be quite difficult. At present, there have been numbers of voice noise reduction algorithms. However, noise reduction effect under strong noisy background is not very well. With a view to improving speech quality, one improved variable-zero-attracting LMS (VZA-LMS) algorithm based on discrete cosine transform (DCT) has been provided in this paper with respect to strong noisy information feature under the oil well. This algorithm has integrated DCT and improved ZA-LMS, utilizing the DCT to improve the signal sparseness at first, one new VZA-LMS algorithm has been put forward thereafter on basis of ZA-LMS algorithm research, by which variance yields amending zero attraction function has been carried out to reduce steady state imbalance deviation. Under the condition of strong noisy background, this algorithm has passed experimental simulation of MATLAB. Compared with current adaptive algorithm, even though calculated amount slightly rising, this new algorithm has better rate of convergence and steady state error, higher SNR( signal to noise ratio) with the verification of better noise elimination ability.