Modeling Irregular Voice in End-to-End Speech Synthesis via Speaker Adaptation | IEEE Conference Publication | IEEE Xplore