BriefGPT.xyz
Aug, 2018
半监督训练以提高端到端语音合成的数据效率
Semi-Supervised Training for Improving Data Efficiency in End-to-End Speech Synthesis
HTML
PDF
Yu-An Chung, Yuxuan Wang, Wei-Ning Hsu, Yu Zhang, RJ Skerry-Ryan
TL;DR
本文提出了一种半监督的训练框架来提高Tacotron数据效率，通过利用大量的公开文本和语音语料库的文本和声学知识，该框架使Tacotron能够使用不到半小时的配对训练数据生成可理解的语音。
Abstract
Although end-to-end
text-to-speech
(TTS) models such as
tacotron
have shown excellent results, they typically require a sizable set of high-quality
pairs for training, which are expensive to collect
→