Music is one of the windows into investigate human cognition. Long-term training will change the musical perception, due to plasticity of the human brain. On the other hand, mental or neural disorders may affect musical perception. The aim of this study is to propose a musical perception experiment and the corresponding auditory evoked potentials (AEP) analysis strategy to distinguish musicians with long-term training, healthy controls and patients with psychiatric disorders. AEPs from 12 schizophrenia patients (defect), 12 musicians (enhancement) and 12 healthy people (regular) invoked by musical intervals and chords were recorded. AEP analysis results show that schizophrenia patients have reduced N1 and P2 amplitudes comparing with the other two groups. Musicians have enhanced N1, P2 and N1-P2 amplitudes. Basing on these observations, a two-layer hierarchical AEP classification method based on the linear discriminate analysis (LDA) is proposed to distinguish these different subject groups. Layer-1 classifies schizophrenia patients with others based on AEP elicited by musical intervals. Layer-2 distinguishes musicians and normal people based on stimuli of consonant intervals and chord. Accuracy of the proposed two-layer method can reach 91.63% through leave-one-out cross validation. Experimental results demonstrate the feasibility of scaling musical perception by analyzing brain AEPs elicited by musical stimuli with different complexity and consonance. These results can be applied to the development of musical training evaluation and psychiatric disorders diagnosis systems in the future.