Improved Wavelet Thresholding Function and Adaptive Thresholding for Noise Reduction
Rong Fu, Jie Zhang, R. Wang, Tao Xu · 2023
In practical scenarios, speech signals are inevitably influenced by the environment during transmission and reception, leading to the presence of noise in the received signal. Consequently, noise reduction plays a vital role in signal processing. Wavelet thresholding algorithms have been widely used for noise reduction due to their simplicity and ease of implementation. In this paper, an improved wavelet thresholding function is proposed for speech signals with strong background noise to retain more useful wavelet coefficients, and the thresholding algorithm is improved to accommodate the deviation of thresholds for different speech signals. The experiment compared three improved wavelet threshold function-based noise reduction algorithms and comprehensively evaluated the noise reduction performance using parameters such as signal-to-noise ratio, mean squared error, waveform similarity, and the filtered waveform. The experimental results indicate that, when dealing with strongly interfering speech signals, the proposed method outperforms other algorithms in preserving valuable information and exhibits significant improvement in noise reduction performance. These findings hold certain research value.