A precise estimation of vocal tract parameters for high quality voice morphing

One of the most recent models for voice conversion is the classical LPC analysis-synthesis model combined with GMM, which aims to separate information from excitation and vocal tract and to learn the transformation rules with statistical methods. However, it does not work well as it is supposed to be due to the inaccuracy of the extracted feature information as well as the overly-smoothed spectral converted by traditional GMM. In this paper, we propose a novel method to solve the problem which is based on the technique of the separation of glottal waveforms and the prediction of the excitations. The final result shows that not only are the transformed vocal tract parameters matching the target one better, but also is the high quality of the synthesized speech preserved.