<p>We introduce a seismic stacking procedure guided by parameters derived from the spectral recomposition of seismic trace spectra. The approach is designed to improve the accuracy of seismic data processing, particularly in the context of offshore hydrocarbon reservoirs. Our methodology begins with an inversion scheme to reconstruct wavelets from seismic traces. This inversion enables the extraction of signal parameters, which are then used to estimate the wavelet shape and temporal position. By correlating these parameters, wavelets associated with the same reflection event are identified, enabling the correction of phase shifts and stretching during stacking. The results obtained show that the proposed stacking procedure, validated on both synthetic and field data, is effective for conventional and converted reflections, demonstrating the potential for improving seismic processing in offshore hydrocarbon exploration and supporting more accurate subsurface characterization.</p>