Related references
Note: Only part of the references are listed.
Proceedings Paper
Audiology & Speech-Language Pathology
Hemlata Tak et al.
Summary: This study utilizes graph attention networks to model relationships for spoofing detection and improve performance, with experiments showing that this approach outperforms traditional methods.
Proceedings Paper
Audiology & Speech-Language Pathology
Xin Wang et al.
Summary: Recent research has focused on back-end neural networks and training criteria for speech spoofing countermeasures. This study offers a comparative perspective on various models and recognizes the potential impact of random initial seed on model performance. Promising techniques, including average pooling and a new hyper-parameter-free loss function, led to the best single model with significantly different statistical performance compared to others.
Proceedings Paper
Audiology & Speech-Language Pathology
Yuxiang Zhang et al.
Summary: The study found that silent intervals affect anti-spoofing measures, VAD operations cause neural networks to lose information on silent segments and lead to severe overfitting. By analyzing different frequency sub-bands, it was discovered that the high-frequency part is the main cause of system overfitting, while the low-frequency part is more robust against known attacks but less accurate.
Proceedings Paper
Audiology & Speech-Language Pathology
Xu Li et al.
Summary: The study introduces a novel channel-wise gated Res2Net (CG-Res2Net) approach, modifying Res2Net to enable a channel-wise gating mechanism in the connection between feature groups, thereby enhancing the detection generalizability of the system.
Proceedings Paper
Acoustics
Hemlata Tak et al.
Summary: This paper presents the first application of RawNet2 to anti-spoofing, showing promising results in detecting various attacks in ASVspoof 2019 evaluation. Results show that RawNet2 systems perform as the second-best in A17 attacks, while the fusion with baseline countermeasures also yields the second-best results reported under ASVspoof 2019 logical access condition.
2021 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP 2021)
(2021)
Proceedings Paper
Acoustics
Jee-weon Jung et al.
Summary: This work introduces a novel back-end framework for speaker verification using graph attention networks, which constructs a graph using segment-wise speaker embeddings and directly outputs a similarity score. The proposed framework includes techniques to enable successful adaptation for speaker verification. Experimental results show consistent improvement of the proposed framework over traditional cosine similarity back-end classifiers.
2021 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP 2021)
(2021)
Proceedings Paper
Acoustics
Xu Li et al.
Summary: By introducing the Res2Net model structure and multi-scale mechanism, the generalizability of the anti-spoofing countermeasure has been improved, and the model size has been reduced. Experimental results show that the performance of Res2Net in the ASVspoof 2019 corpus is significantly better than other models, especially excelling in physical access and logical access scenarios.
2021 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP 2021)
(2021)
Proceedings Paper
Acoustics
Anwei Luo et al.
Summary: The study introduced a capsule network to enhance the generalization of audio anti-spoofing systems, to detect fake audios synthesized by advanced methods and combat various attacks, including text-to-speech and voice conversion attacks. The results demonstrated that this approach is also highly capable of detecting replay attacks.
2021 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP 2021)
(2021)
Article
Engineering, Electrical & Electronic
Guang Hua et al.
Summary: This paper introduces a new synthetic speech detection approach, the TSSDNet model, which utilizes end-to-end DNN and eliminates the need for hand-crafted feature extraction. Experimental results show significant performance improvement, demonstrating the potential and advantages of this model in synthetic speech detection.
IEEE SIGNAL PROCESSING LETTERS
(2021)
Article
Engineering, Electrical & Electronic
You Zhang et al.
Summary: This study introduces an anti-spoofing system using one-class learning to detect unknown synthetic voice spoofing attacks, achieving excellent results without relying on data augmentation methods.
IEEE SIGNAL PROCESSING LETTERS
(2021)
Article
Chemistry, Multidisciplinary
Hye-jin Shim et al.
APPLIED SCIENCES-BASEL
(2020)
Proceedings Paper
Computer Science, Theory & Methods
Xiao Wang et al.
WEB CONFERENCE 2019: PROCEEDINGS OF THE WORLD WIDE WEB CONFERENCE (WWW 2019)
(2019)
Article
Computer Science, Theory & Methods
Xiang Wu et al.
IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY
(2018)
Proceedings Paper
Computer Science, Artificial Intelligence
Galina Lavrentyeva et al.
18TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2017), VOLS 1-6: SITUATED INTERACTION
(2017)
Proceedings Paper
Computer Science, Artificial Intelligence
Tomi Kinnunen et al.
18TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2017), VOLS 1-6: SITUATED INTERACTION
(2017)
Proceedings Paper
Acoustics
Kaavya Sriskandaraja et al.
17TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2016), VOLS 1-5: UNDERSTANDING SPEECH PROCESSING IN HUMANS AND MACHINES
(2016)
