Past studies of wavelet technologies for gamma spectral analysis essentially focused on direct fitting of raw gamma spectra, but these studies often failed to produce new benefits for operational adaptation of wavelet analysis. This paper presents a modified wavelet approach with the objective being detecting only the nuclides that do not exist in the environmental background. With this operational objective, wavelet analysis is applied to the background-subtracted count-rate spectra. A preliminary comparison study suggests that this background subtraction - wavelet-fitting process - is independent of the detector type and background radiation and is capable of improving the wavelet peak detection probabilities as compared with earlier published results.