Statistical Text-to-Speech Synthesis Based on Segment-Wise Representation With a Norm Constraint | IEEE Journals & Magazine | IEEE Xplore