Abstract: In this paper wavelet transform is applied to pitch detection of Chinese trisyllabic word with characteristic of continuous speech, a new algorithm for pitch detection of speech signals using wavelet transform is proposed for the question appearing in pitch extraction. It is made up of the following processes: calculating discrete wavelet transform, selecting pitch points and determining the first pitch point based on voting strategy, checking pitch, modifying abnormal pitch point, processing pitch miss point at the beginning and end of a frame of speech signals, relocation of pitch point based on voting strategy. It has been found that this algorithm is valid, and very suits to process continuous speech. But it needs much time for computation, so it is not suit for real-time system.