Towards an Improved Modeling of the Glottal Source in Statistical Parametric Speech Synthesis (2008)
João P. Cabral, Steve Renals, Korin Richmond, Junichi Yamagishi
This paper proposes the use of the Liljencrants-Fant model (LFmodel) to represent the glottal source signal in HMM-based speech synthesis systems. These systems generally use a pulse train to model...