A GEOMETRICAL FUZZY CLUSTERING BASED SOLUTION TO GLOTTAL WAVE ESTIMATION

Accurate estimation of the glottal waveform (GW) is required for purpo ses such as natural speech synthesis, speaker recognition, physiologic al speech processing, etc. Most methods available for GW estimation ar e based on inverse filtering of the speech signal through the vocal tr act, and they all suffer from inaccuracies due to incorrect assumption s. The method for GW estimation developed in the present study is base d on fuzzy clustering of quasi-linear geometrical substructures, repre sented within the signal shifts hyperspace. Algorithms for estimation of the driving function to the vocal tract are presented and evaluated on simulated and real data. Comparison of the fuzzy clustering-based method with the PSIAIF and Wong's closed-phase algorithms shows that t he present method is superior with respect to both the GW estimation a nd determination of GW event time instants. (C) 1998 Acoustical Societ y of America. [S0001-4966(98)04411-7].