In an image forming apparatus, a controller allows a sound output device to output a starting sound and an ending sound at a start and end of diagnostic image formation processing, respectively, determines whether abnormal noise has occurred by identifying data ranging from the starting sound to the ending sound in an diagnostic spectrogram as a comparison target for data ranging from the starting sound to the ending sound in a normal spectrogram to match both the spectrograms in terms of width in a time-axis direction, and identifies a source of the abnormal noise by identifying the data ranging from the starting sound to the ending sound in the diagnostic spectrogram as a comparison target for data in ranging from a timing of start to a timing of end in a timing chart to match the diagnostic spectrogram with the timing chart in terms of width in the time-axis direction.