The current noise reduction method only uses the filter to deal with a certain kind of special noise to achieve the purpose of noise reduction. When applied to electronic music noise reduction, the noise reduction effect is poor, low efficiency, and affect the human ear sensory experience of music. Aiming at the above defects, a method of electronic music signal denoising based on Naive Bayes is studied. After digital preprocessing of electronic music signal, spectrum characteristics of electronic music are analyzed from the angle of human ear. After node filtering of wavelet packet decomposition tree by Naive Bayes model, the electronic music signal is decomposed and reconstructed by the optimal wavelet packet decomposition tree to complete noise reduction. In the noise reduction experiment, the signal-to-noise ratio of signal processed by naive Bayesian method is reduced by about 2∼5dB, and the noise reduction efficiency is improved by about 32.87%, and the noise reduction effect is better.