Train from scratch: Single-stage joint training of speech separation and recognition

doi:10.1016/j.csl.2022.101387

	Train from scratch: Single-stage joint training of speech separation and recognition
	Shi, Jing1 ; Chang, Xuankai 2; Watanabe, Shinji 2; Xu, Bo1,3
发表期刊	COMPUTER SPEECH AND LANGUAGE
ISSN	0885-2308
	2022-11-01
卷号	76 页码:15
通讯作者	Watanabe, Shinji(shinjiw@ieee.org)
摘要	Multi-speaker speech separation and recognition gains much attention in the speech community recently. Previously, most studies train the front-end separation module and back-end recognition module individually. The two modules after training are combined together either with a hybrid structure or by fine-tuning the resulting model. In this work, we present a unified and flexible multi-speaker end-to-end ASR model. In contrast to previous studies, our proposed model is trained from scratch with a complete single stage, rather than multiple training stages based on pre-training and the following fine-tuning. Our model can deal with either single channel or multi-channel speech input. Moreover, the proposed model can be trained with or without the clean source speech references. We evaluate the proposed model on the WSJ02mix dataset in both single-channel and spatialized multi-channel conditions. The experiments demonstrate that the proposed methods can improve the performance of the end-to-end model in recognizing the separated streams without much degradation in speech separation, achieving a new state-of-the-art in the WSJ0-2mix dataset. Moreover, we systematically assess the impact of various features for the success of the joint-training model and will release all our codes, which may provide a new guidance for the integration of front-end and back-end towards complex auditory scenes.
关键词	Cocktail party problem Speech separation Multi-speaker speech recognition End-to-end Joint-training
DOI	10.1016/j.csl.2022.101387
关键词[WOS]	DOMAIN AUDIO SEPARATION ; NEURAL-NETWORKS ; END
收录类别	SCI
语种	英语
WOS研究方向	Computer Science
WOS类目	Computer Science, Artificial Intelligence
WOS记录号	WOS:000798734700002
出版者	ACADEMIC PRESS LTD- ELSEVIER SCIENCE LTD
引用统计	被引频次：3[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/49507
专题	复杂系统认知与决策实验室_听觉模型与认知计算
通讯作者	Watanabe, Shinji
作者单位	1.Chinese Acad Sci CASIA, Inst Automat, Beijing, Peoples R China 2.Johns Hopkins Univ, Ctr Language & Speech Proc, Baltimore, MD 21218 USA 3.Univ Chinese Acad Sci, Beijing, Peoples R China
第一作者单位	中国科学院自动化研究所
推荐引用方式 GB/T 7714	Shi, Jing,Chang, Xuankai,Watanabe, Shinji,et al. Train from scratch: Single-stage joint training of speech separation and recognition[J]. COMPUTER SPEECH AND LANGUAGE,2022,76:15.
APA	Shi, Jing,Chang, Xuankai,Watanabe, Shinji,&Xu, Bo.(2022).Train from scratch: Single-stage joint training of speech separation and recognition.COMPUTER SPEECH AND LANGUAGE,76,15.
MLA	Shi, Jing,et al."Train from scratch: Single-stage joint training of speech separation and recognition".COMPUTER SPEECH AND LANGUAGE 76(2022):15.