Laser sound detection technology is an important research direction in the field of voice detection, but this detection technology is vulnerable to noise interferences, which are mainly caused by background light and atmospheric turbulence. The noise suppression of detection signal is the key of laser sound detection technology. Therefore, an improved threshold function is proposed. By adjusting parameters, the deviation between the wavelet coefficients and the original wavelet coefficients can be changed, and the feature information of the original signals can be preserved as much as possible. In laboratory environment, the effectiveness of wavelet threshold denoising method based on the improved threshold function is verified by experiments. After denoising, the noise is effectively removed.