In practice, fluctuations around the nominal operating speed often result in vibration spectra of gearboxes to appear smeared. Order tracking technique can be employed to remove the spectral smearing of such signals. However, this technique requires simultaneous measurements of vibration and rotational speed, which is often unavailable in industrial settings. Thus, a new diagnostic algorithm without a tachometer is proposed in this paper for monitoring of gear faults. This new tacho-less method uses fast dynamic time warping (FDTW) algorithm to remove the spectral smearing, is based on the introduction of an estimated reference signal having the same frequency as the nominal shaft rotational frequency. The effectiveness of this method is validated using both simulated signals from a fixed axis gear pair and experimental signals from a planetary gearbox test rig. The results demonstrate the ability of the proposed approach in extracting fault information from gearbox vibration signals under fluctuating speed conditions.