Improved Variable Step Size Regularized NLMS-Based Algorithm for Speech Enhancement

To minimize the noise level in speech signals, many adaptive filters, such as LMS and NLMS, are utilized to reach the steady state. The filter weights are adapted based on specific functions to enhance the signal to noise ratio (SNR) of the system output with a faster convergence speed. This paper proposes a new LMS-based variable step size algorithm to reduce the noise level in the corrupted speech. The new algorithm is compared with other variable step size algorithms in speech enhancement. The results show that the newproposed algorithm improves the SNR at the filter output. Moreover, it has fixed SNR improvement over a wide range of different input SNR values. Over this range, the proposed algorithm has more stable performance measures and achieves higher convergence speed and lower steady state error. It is also able to get very small misalignment values between the filter weights and the targeting channel.