Nội dung này: Investigations on speaker adaptation using a continuous vocoder within recurrent neural network based text-to-speech synthesis