Spatial reconstructed local attention Res2Net with F0 subband for fake speech detection

doi:10.1016/j.neunet.2024.106320

CORC > 自动化研究所 > 中国科学院自动化研究所 > 模式识别国家重点实验室 > 模式分析与学习团队

	Spatial reconstructed local attention Res2Net with F0 subband for fake speech detection
	Fan, Cunhang 4; Xue, Jun 4; Tao, Jianhua 3; Yi, Jiangyan 2; Wang, Chenglong 2; Zheng, Chengshi 1; Lv, Zhao 4
刊名	NEURAL NETWORKS
	2024-07-01
卷号	175 页码:11
关键词	ASVspoof Fake speech detection Fundamental frequency Res2Net
ISSN号	0893-6080
DOI	10.1016/j.neunet.2024.106320
通讯作者	Fan, Cunhang(cunhang.fan@ahu.edu.cn) ; Tao, Jianhua(jhtao@tsinghua.edu.cn)
英文摘要	The rhythm of bonafide speech is often difficult to replicate, which causes that the fundamental frequency (F0) of synthetic speech is significantly different from that of real speech. It is expected that the F0 feature contains the discriminative information for the fake speech detection (FSD) task. In this paper, we propose a novel F0 subband for FSD. In addition, to effectively model the F0 subband so as to improve the performance of FSD, the spatial reconstructed local attention Res2Net (SR-LA Res2Net) is proposed. Specifically, Res2Net is used as a backbone network to obtain multiscale information, and enhanced with a spatial reconstruction mechanism to avoid losing important information when the channel group is constantly superimposed. In addition, local attention is designed to make the model focus on the local information of the F0 subband. Experimental results on the ASVspoof 2019 LA dataset show that our proposed method obtains an equal error rate (EER) of 0.47% and a minimum tandem detection cost function (min t-DCF) of 0.0159, achieving the state-of-the-art performance among all of the single systems.
资助项目	STI[2021ZD0201500] ; National Natural Science Foundation of China (NSFC)[62201002] ; National Natural Science Foundation of China (NSFC)[62322120] ; Excellent Youth Foundation of Anhui Scientific Committee, China[208085J05] ; Special Fund for Key Program of Science and Technology of Anhui Province, China[202203a07020008] ; Open Research Projects of Zhejiang Lab, China[2021KH0AB06]
WOS研究方向	Computer Science ; Neurosciences & Neurology
语种	英语
出版者	PERGAMON-ELSEVIER SCIENCE LTD
WOS记录号	WOS:001234206200001
资助机构	STI ; National Natural Science Foundation of China (NSFC) ; Excellent Youth Foundation of Anhui Scientific Committee, China ; Special Fund for Key Program of Science and Technology of Anhui Province, China ; Open Research Projects of Zhejiang Lab, China
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/58487]
专题	自动化研究所_模式识别国家重点实验室_模式分析与学习团队
通讯作者	Fan, Cunhang; Tao, Jianhua
作者单位	1.Chinese Acad Sci, Inst Acoust, Key Lab Noise & Vibrat Res, Beijing 100190, Peoples R China 2.Chinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100190, Peoples R China 3.Tsinghua Univ, Dept Automat, Beijing 100190, Peoples R China 4.Anhui Univ, Sch Comp Sci & Technol, Anhui Prov Key Lab Multimodal Cognit Computat, Hefei 230601, Peoples R China
推荐引用方式 GB/T 7714	Fan, Cunhang,Xue, Jun,Tao, Jianhua,et al. Spatial reconstructed local attention Res2Net with F0 subband for fake speech detection[J]. NEURAL NETWORKS,2024,175:11.
APA	Fan, Cunhang.,Xue, Jun.,Tao, Jianhua.,Yi, Jiangyan.,Wang, Chenglong.,...&Lv, Zhao.(2024).Spatial reconstructed local attention Res2Net with F0 subband for fake speech detection.NEURAL NETWORKS,175,11.
MLA	Fan, Cunhang,et al."Spatial reconstructed local attention Res2Net with F0 subband for fake speech detection".NEURAL NETWORKS 175(2024):11.