Utterance-Level Sequential Modeling for Deep Gaussian Process Based Speech Synthesis Using Simple Recurrent Unit | IEEE Conference Publication | IEEE Xplore