An improved pitch extraction algorithm for speech processing

Summary form only given. In this paper, we propose an improved pitch extraction algorithm for speech processing. Base on the original autocorrelation function algorithm, we use three measures to rectify the weight and proportion of right pitch values in the search space. The proposed algorithm is evaluated on the Keele database and FDA database. Experimental results show that the proposed method has a great increase in performance over the other algorithms. Compared with the original algorithm, voiced error and pitch tract error have a relative reduction of 28.74% and 5.53%, respectively. The proposed method is more suitable for speech processing.