A gearing system emits inconsistent noises from the impact of gear teeth when defects are present, but it is not easy for a noise inspector on a production line to distinguish defective products objectively. Since customers constantly complain about various noises from geared motors, it is desirable to devise an analytical technique to classify motors. However, it is difficult to separate inconsistent noises due to defective gears from the overall noise produced by a geared motor using a general signal processing method such as a FFT because low frequency impulse signals have a tendency not to appear in the frequency domain. In this paper, we propose a method that can be used to obtain more objective estimates and measurements of inconsistent noises from a gearing system. The proposed method makes use of the cepstrum domain with an applied autocorrelation and comb lifter, followed by a domain inversion.