In hybrid mode, a certain electric hybrid bus would produce abnormal squealing noise during the braking process. Combined with the subjective judgement of the noise source, the noise signal from the attachment location of transmission and the permanent magnet motor is collected and its time–frequency characteristics is analyzed based on the complex Morlet wavelet algorithm. In addition, the time–frequency domain information of abnormal noise is determined with the help of filtering and sound playback technology. It is found that abnormal noise is related to vehicle speed. The abnormal noise frequencies at various initial braking velocities were extracted in comparison to the biting sound frequencies of connected bevel gears between the gearbox and permanent magnet motor. It is found that the two frequencies are very close. Therefore, it is basically concluded that the abnormal noise of this bus is generated by the biting transmission of bevel gears. The abnormal noise was basically eliminated in the real vehicle test after the bevel gear pairs were carefully matched and modified before loading which further confirms the noise source. From above, it shows that the time–frequency analysis method based on complex Morlet wavelet provide an accurate method to identify the brake noise source.