In this paper, the signal denoising method based on wavelet threshold is studied, and the threshold function design is discussed. The traditional hard threshold function method has a larger variance in de-noising results, and the soft threshold function method has smooth results, but the soft threshold function method also reduces the amplitude of the wavelet coefficients and the denosing result may appear a big deviation. Based on this, two improved threshold functions are proposed in this paper, and the denoising effects of the improved methods are obvious, and the SNR is greatly improved.