Pitch period detection using second generation wavelet transform

Pitch period estimation is one of the important issues in speech processing. Wavelet transform has been used for pitch detection in recent research. In this paper a new modified algorithm based on second generation wavelet transform and cubed speech signal for pitch period determination, in conjunction with other characteristics enhancement methods, is presented. This algorithm demands less computation in comparison with those using classic wavelet methods for the same purpose. The proposed algorithm has been evaluated both with real and synthesized speech in noisy and ideal circumstances. The results indicate that the proposed method provides a good estimation of pitch period and its sensitivity to noise is fairly good.