To address the difficulty in identifying liquid-level reflection points arising from the susceptibility of dynamic liquid-level echoes to interference from downhole operating conditions during drilling, this study proposes a method for identifying and extracting dynamic liquid-level signals in the time–frequency (T–F) plane using the synchrosqueezing generalized phase-shifting S-transform (SS-GPST). The method introduces adjustable parameters to regulate T–F resolution and phase-shift correction to improve phase consistency between the S-transform and synchrosqueezing reassignment, thereby obtaining high-resolution T–F distributions of dynamic liquid-level signals. On this basis, two-dimensional Gaussian filtering is applied in the T–F plane to smooth and suppress background noise. The T–F region containing the liquid-level reflection signal is then extracted for inverse-transform reconstruction, and the liquid-level reflection point and its occurrence time are determined from the position at which a pronounced abrupt change first occurs in the reconstructed waveform. The results show that, compared with the T–F distributions obtained using the synchrosqueezing S-transform (SS-ST), synchrosqueezing phase-shifting S-transform (SS-PST), and synchrosqueezing generalized S-transform (SS-GST), the SS-GPST achieves higher T–F resolution and preserves more local details. The measurement error of the SS-GPST does not exceed 1.2 m, and its relative error remains within 0.083%, confirming its ability to identify and extract dynamic liquid-level signals more accurately under complex interference conditions.