SPEECH ENHANCEMENT BY HARMONIC MODELING VIA MAP PITCH
Joseph Tabrikian, Shlomo Dubnov, Yulya Dickalov, Beer Sheva · 2002
In this paper we pres ent a p rocedure for estimating the parame ters of speech signals that are contaminated by high level of noise. The proposed estimation method is developed by assuming a harmonic model for the voiced frame hypothesis. A Maximum A-posteriori Probability tracking method is developed for esti mating time-varying pitch. Signal reconstruction is achieved by projecting the signal onto the subspace of harmonic signals with the optimal estimates of the fundamental frequency. The per formance of the proposed method is evaluated and compared to other existing methods using a large pitch detection database. It is shown that the proposed method for pitch estimation is more robust and much more accurate in terms of mean-square-error and gross error rate, in comparison to other existing methods, speci ally at ultra low signal-to-n oise ratios (as low as -15 dB). Ex amples of speech reconstruction/enhan cement are also presented in the paper.