Abstract: Representation learning is one of the fundamental issues in modeling articulatory-based speech synthesis using target-driven models. This paper proposes a computational strategy for learning ...