Emotionally-Enhanced Text-to-Speech  Synthesis Using Convolutional Neural Networks