Accurate calculation of water content is of great significance for mastering the production capacity, channeling, and water flooding of oil wells. This article proposes an optimized empirical wavelet transform (EWT) and frequency wavenumber spectrum (FK) analysis based method for measuring water content, which addresses the issue of noise interference affecting the measurement of sound velocity and resulting in high uncertainty in water content measurement. The experimental results show that compared with WT, EMD-WT, and CEEMDAN-WT denoising methods, this method exhibits excellent capabilities in both noise filtering and maintaining the original signal waveform. The final measured curve of the relationship between sound velocity and water content has a fitting degree of 94.36%, which is superior to other methods.