Neuron can be excited and inhibited by filtered signals. The filtering properties of neural networks have a huge impact on memory, learning, and disease. In this paper, the filtering properties of Hodgkin-Huxley neuron to different time-scale signals are investigated. It is found that the neuronal filtering property depends on the locking relationship between the signal's frequency band and the natural frequency of neuron. The natural firing frequency is a combination of the fundamental component and the various level harmonic components. The response of neuron to the filtered signal is related to the amplitude of the harmonic components. Neuron responds better to the low-frequency signals than the high-frequency signals because of the reduction in the harmonic component amplitude. The filtering ability of neuron can be modulated by the excitation level, and is stronger around the excitation threshold. Our results might provide novel insights into the filtering properties of neural networks and guide the construction of artificial neural networks.