The pitch detection is very important in automatic music transcription in general. In this paper, an algorithm of detecting pitch period from Erhu music signal based on Hilbert-Huang transform (HHT) is proposed. Traditional windowing methods generally employ overlapping windows and have an assumption of stationary pitch period within a window. In contrast, HHT shows no limitations on window selection and allows pitch period changing within windows. And it also can be used to monitor the variation of the pitch. In this article, description of the principles and characteristics of HHT and comparison between HHT method and Short Time Fourier Transform (STFT) method are presented. To validate the proposed method, the international standard pitch is used. It is observed from the experimental results that the proposed algorithm is more precise than STFT. The results show that the variation of the pitch period can be accurately detected. It indicates that the proposed method can be applied into the emotion analysis of the erhu performer. This demonstrates Hilbert-Huang transform can be applied to detect pitch from erhu music signal.