<p>Free Space Optical (FSO) Communication, as a novel communication method, has advantages in terms of interference resistance and confidentiality compared to radio frequency (RF) technology. Since the spectrum of voice signals in FSO communication is confined to a specific range, thereby inherently limiting the range of frequencies that can be processed in the frequency domain. To enhance the noise resistance of voice signals in communications, we employed a method based on optimized weight spectra to identify and remove noise from signals. The sample signal is transformed into the frequency domain via Fourier transform, where noise is removed by applying an optimized weighting spectrum. Subsequently, the signal is reconstructed through inverse Fourier transform. The designed algorithm analyzed the bit error rate (BER) under different signal-to-noise ratios through simulation. Using the developed optical communication equipment at a distance of 80 m outdoors to test the BER of different modulation methods, the optimized weighted spectrum has a significantly lower BER, which is lower than 9.31 ‰. In this study, compared to the method of directly retaining effective frequency components, the BER significantly decreases, and the ability of noise reduction and anti-interference is notably improved. The use of new modulation techniques enhanced the reliability of voice information transmission, demonstrated potential applicability, and improved the environmental adaptability of FSO communication systems.</p>