Due to the size of the micro motor is small, its vibration signal cannot be collected by conventional acceleration sensor, and the specific fault type of the micro motor does not need to be diagnosed, only needs to judge whether the fault exists. Therefore, noise detection is usually adopted in the early stage of fault detection of micro motor. Based on this detection method, a fault diagnosis method for micro motor was proposed based on sound signal.Aimed at the characteristics that sound signal has high signal-to-noise ratio and is easy to be affected by environment,the maximum correlated kurtosis deconvolution (MCKD)-wavelet threshold de-noising method was used to enhance the periodic impact component in the sound signal and filter out the environmental noise. The envelope and envelope spectrum of the signal were obtained by using Hilbert transform.According to the shape of the envelope and the frequency corresponding to the peak value of the envelope spectrum, the fault diagnosis for the micro motor was realized