Rolling bearings are critical components in modern mechanical equipment, and the health of bearings is closely related to the normal operation of equipment. Operating machinery under complex conditions may generate compound fault signals containing various periodic impacts. Therefore, separating different types of fault components has gradually become a focus of research. In this paper, a weighted spectrum editing method based on amplitude mapping is proposed. The constructed hyperbolic tangent Gaussian function is used to edit the original spectrum, obtaining a series of weighted spectra with different fluctuation levels. These weighted spectra are then applied to the original spectrum to achieve amplitude editing of the signal spectrum. Subsequently, the edited spectrum is combined with the original phase using Fourier inverse transformation to obtain a series of reconstructed signals. In the envelope spectrum of the reconstructed signal, different types of fault features are separated to achieve the diagnosis of compound faults. Simulation and experimental signals demonstrate that this method can not only extract different periodic components from compound fault signals and suppress noise interference but also exhibit good anti-aliasing characteristics compared to traditional frequency band division and filtering methods.