Fig. 1From: A multi-task learning speech synthesis optimization method based on CWT: a case study of Tacotron2Basic framework for multi-task learningBack to article page