Sandwich structures with polyurethane foam core and glass fiberreinforced polymer facesheets with three orientations were investigated experimentally and numerically under three-point bending tests at 80C, at a relatively low load level associated with a linear viscoelastic response. Off-the-shelf piezoelectric transducers were inserted inside one of the facesheets and were interrogated in pitch-catch at low ultrasonic frequencies during testing. The objective of this article is to investigate ability and sensitivity of the embedded transducers to detect creep deformation. The denoised received waveforms were analyzed in the time domain, where guided wave speeds were found to exhibit a drop due to temperature changes (most significant in the sandwich samples with off-axis orientation), followed by an increase, eventually reaching an asymptotic value. The waveforms were also processed in the joint frequencytime domain, with a novel signal processing technique built upon Gabor wavelet transforms and their contour lines. It is shown that this wavelet contour technique indirectly captures the trend of physically measured displacements and can differentiate among the three different fiber orientations in the facesheets, and among room temperature and 80C. This technique has the potential to effectively track creep time-dependent response and life performance in smart sandwich composites.