Spoken Content and Voice Factorization for Few-shot Speaker Adaptation

CASIA OpenIR > 多模态人工智能系统全国重点实验室 > 模式分析与学习

	Spoken Content and Voice Factorization for Few-shot Speaker Adaptation
	Wang T(汪涛)
	2020-04
会议名称	INTERSPEECH 2020
会议日期	2020
会议地点	Online
摘要	The low similarity and naturalness of synthesized speech remain a challenging problem for speaker adaptation with few resources. Since the acoustic model is too complex to interpret, overfifitting will occur when training with few data. To prevent the model from overfifitting, this paper proposes a novel speaker adaptation framework that decomposes the parameter space of the end-to-end acoustic model into two parts, with the one on predicting spoken content and the other on modeling speaker’s voice. The spoken content is represented by phone posteriorgram (PPG) which is speaker independent. By adapting the two sub-modules separately, the overfifitting can be alleviated effectively. Moreover, we propose two different adaptation strategies based on whether the data has text annotation. In this way, speaker adaptation can also be performed without text annotations. Experimental results confifirm the adaptability of our proposed method of factorizating spoken content and voice. Listening tests demonstrate that our proposed method can achieve better performance with just 10 sentences than speaker adaptation conducted on Tacotron in terms of naturalness and speaker similarity.
七大方向——子方向分类	智能交互
国重实验室规划方向分类	语音语言处理
是否有论文关联数据集需要存交	否
文献类型	会议论文
条目标识符	http://ir.ia.ac.cn/handle/173211/52367
专题	多模态人工智能系统全国重点实验室_模式分析与学习
通讯作者	Wang T(汪涛)
作者单位	Institute of Automation, Chinese Academy of Sciences
推荐引用方式 GB/T 7714	Wang T. Spoken Content and Voice Factorization for Few-shot Speaker Adaptation[C],2020.

条目包含的文件		下载所有文件
文件名称/大小	文献类型	版本类型	开放类型	使用许可
Spoken Content and V（1514KB）	会议论文		开放获取	CC BY-NC-SA	浏览下载