Potential of Gaia XP spectra in red giant star asteroseismology: A deep-learning approach

Red giants are key tracers of stellar evolution and Galactic structure, and their asteroseismic properties — particularly the large frequency separation (Δν), the frequency of maximum oscillation power (ν_ max ), and the dipole-mode period spacing (ΔΠ_1) — provide direct insight into their internal structure, masses, and evolutionary states. Until now, seismic inferences on large stellar samples have relied primarily on high-quality light curves from missions such as and or on moderate-resolution spectroscopy (LAMOST : mathcal Kepler TESS R and APOGEE : mathcal R that clearly preserve information correlated with these seismic quantities. With Gaia XP spectra (mathcal R the possibility arises to obtain asteroseismic measurements of orders of magnitude more stars, despite the much lower spectral resolution. Our goal is to assess whether XP spectra retain enough information to enable reliable seismic inference for red giants. We developed hybrid convolutional neural network (CNN)--long short-term memory (LSTM) models trained on red giants with seismic parameters measured from photometry. The networks learn the subtle spectral signatures — imprinted through global stellar properties — that correlate with Δν, ν_ Kepler max , and ΔΠ_1. The models recover all three global asteroseismic parameters from Gaia XP spectra with accuracies comparable to results based on moderate-resolution surveys such as LAMOST, demonstrating that even low-resolution spectrophotometry carries sufficient information for seismic prediction. Saliency analysis reveals wavelength regions most strongly associated with seismic sensitivity and highlights the physically distinct spectral behavior between RGB and RC stars. Applying our models to Gaia DR3 yielded seismic predictions for more than 2.5 million bright red giants, which allows for population-level asteroseismic studies on an unprecedented scale. We also identified a small subset of low-Δν red clump candidates that show unusual spectral-seismic correlations, offering new avenues for investigating evolved stellar populations.