Lip Motion Synthesis for Speech Animation Using Active Shape Model

摘要

Visual speech synthesis will surely make the machine more human-friendly. The modelling and synthesis of the movements of visible articulators such as lips are predominant components in the production pipeline for talking face synthesis, especially in speech animation. Lip being the most dynamic visible articulator lip modelling is attempted as a separate module in many works. Set of Phonemes and corresponding dynamics of visible articulators demands language specific exploration. This work attempts to model the lip shapes of Malayalam Phoneme utterances using the modelling scheme used in Active Shape Model (ASM). The movements corresponding to phoneme transition are created by forming trajectories in the model coefficient space. Automation of lip movements for speech animation will help to develop cost effective multimedia applications in the domain of education and entertainment.