What to remember: Self-adaptive continual learning for audio deepfake detection
Open accessAbstract
The rapid evolution of speech synthesis and voice conversion has raised substantial concerns due to the potential misuse of such technology, prompting a pressing need for effective audio deepfake detection mechanisms. Existing detection models have shown remarkable success in discriminating known deepfake audio, but struggle when encountering new attack types. To address this challenge, one of the emergent effective approaches is continual learning. In this paper, we propose a continual learning approach called Radian Weight Modification (RWM) for audio deepfake detection. The fundamental concept underlying RWM involves categorizing all classes into two groups: those with compact feature distributions across tasks, such as genuine audio, and those with more spread-out distributions, like various types of fake audio. These distinctions are quantified by means of the in-class cosine distance, which subsequently serves as the basis for RWM to introduce a trainable gradient modification direction for distinct data types. Experimental evaluations against mainstream continual learning methods reveal the superiority of RWM in terms of knowledge acquisition and mitigating forgetting in audio deepfake detection. Furthermore, RWM's applicability extends beyond audio deepfake detection, demonstrating its potential significance in diverse machine learning domains such as image recognition.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Radian Weight Modification (RWM) is a novel architectural approach for continual learning that categorizes classes based on feature distribution similarity.
“The authors of [71] developed the Radian Weight Modification (RWM) method, a novel architectural approach for continual learning that categorizes classes based on feature distribution similarity.”
From Deepfake Media Forensics: Status and Future Challenges (2025) · cites this paper as [71]
The Abstract and Conclusion identify RWM as a proposed continual-learning method that groups classes by feature-distribution similarity. However, the Proposed Method describes it as an algorithm for modifying gradient directions, not as an architectural approach; that characterization is not supported by our copy. (LLM, conf=0.85)
Experimental — accuracy not yet measuredThe feature distribution of real speech across different tasks is relatively compact, whereas the distribution of fake speech features is more dispersed.
“Zhang et al. [161] observed that the feature distribution of real speech across different tasks is relatively compact, whereas the distribution of fake speech features is more dispersed.”
From Audio Deepfake Detection: What Has Been Achieved and What Lies Ahead (2025) · cites this paper as [161]
RWM significantly outperforms mainstream continual learning methods in terms of knowledge memorisation.
“Experimental results demonstrated that RWM significantly outperforms mainstream continual learning methods in terms of knowledge memorisation.”
From Audio Deepfake Detection: What Has Been Achieved and What Lies Ahead (2025) · cites this paper as [161]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · see citing works
Self-citation(2008) · Journal of Machine Learning Research · Cited by 33,051 · see citing works
Overcoming catastrophic forgetting in neural networks
Kirkpatrick, J., Pascanu, R., Rabinowitz, N., Veness, J., Desjardins, G., Rusu, A. A., Milan, K., Quan, J., Ramalho, T., Grabska-Barwinska, A., et al. (2017) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1611835114 · Cited by 7,899 · see citing works
Generalized Inverses: Theory and Applications
Rohatgi, V. K., Ben-Israel, A., & Greville, T. N. E. (1976) · International Statistical Review · doi:10.2307/1403291 · Cited by 4,568 · see citing works
Li, Z., & Hoiem, D. (2018) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2017.2773081 · Cited by 4,202 · see citing works
Continual lifelong learning with neural networks: A review
Parisi, G. I., Kemker, R., Part, J. L., Kanan, C., & Wermter, S. (2019) · Neural Networks · doi:10.1016/j.neunet.2019.01.012 · Cited by 3,117 · see citing works
Wittenmark, B. (1993) · Automatica · doi:10.1016/0005-1098(93)90162-m · Cited by 3,067 · see citing works
wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations
Baevski, A., Zhou, H., Mohamed, A., & Auli, M. (2020) · arXiv · doi:10.48550/arxiv.2006.11477 · Cited by 2,369 · see citing works
(2007) · Cited by 937 · see citing works
Efficient Lifelong Learning with A-GEM
Chaudhry, A., Ranzato, M., Rohrbach, M., & Elhoseiny, M. (2018) · arXiv · doi:10.48550/arxiv.1812.00420 · Cited by 852 · see citing works
ASVspoof 2019: Future Horizons in Spoofed and Fake Audio Detection
Todisco, M., Wang, X., Vestman, V., Sahidullah, M., Delgado, H., Nautsch, A., Yamagishi, J., Evans, N., Kinnunen, T., & Lee, K. A. (2019) · arXiv · doi:10.48550/arxiv.1904.05441 · Cited by 688 · see citing works
ASVspoof 2019: Future Horizons in Spoofed and Fake Audio Detection
Todisco, M., Wang, X., Vestman, V., Sahidullah, M., Delgado, H., Nautsch, A., Yamagishi, J., Evans, N., Kinnunen, T. H., & Lee, K. A. (2019) · Interspeech 2019 · doi:10.21437/interspeech.2019-2249 · Cited by 687 · see citing works
Unsupervised Cross-Lingual Representation Learning for Speech Recognition
Conneau, A., Baevski, A., Collobert, R., Mohamed, A., & Auli, M. (2021) · Interspeech 2021 · doi:10.21437/interspeech.2021-329 · Cited by 637 · see citing works
Style Tokens: Unsupervised Style Modeling, Control and Transfer in End-to-End Speech SynthesisPreprint
Wang, Y., Stanton, D., Zhang, Y., Skerry-Ryan, R., Battenberg, E., Shor, J., Xiao, Y., Ren, F., Ye, J., & Saurous, R. A. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1803.09017 · Cited by 628 · see citing works
The ASVspoof 2017 challenge: assessing the limits of replay spoofing attack detection
Kinnunen, T., Sahidullah, M., Delgado, H., Todisco, M., Evans, N., Yamagishi, J., & Lee, K. A. (2017) · Interspeech 2017 · doi:10.21437/interspeech.2017-1111 · Cited by 535 · see citing works
GDumb: A Simple Approach that Questions Our Progress in Continual Learning
Prabhu, A., Torr, P. H. S., & Dokania, P. K. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58536-5_31 · Cited by 497 · see citing works
ASVspoof 2015: the first automatic speaker verification spoofing and countermeasures challenge
Wu, Z., Kinnunen, T., Evans, N., Yamagishi, J., Hanilçi, C., Sahidullah, M., & Sizov, A. (2015) · Interspeech 2015 · doi:10.21437/interspeech.2015-462 · Cited by 484 · see citing works
Does Audio Deepfake Detection Generalize?Preprint
Müller, N. M., Czempin, P., Dieckmann, F., et al. (2022) · Interspeech · doi:10.48550/arxiv.2203.16263 · Cited by 401 · see citing works
A comparison of features for synthetic speech detection
Sahidullah, M., Kinnunen, T., & Hanilçi, C. (2015) · Interspeech 2015 · doi:10.21437/interspeech.2015-472 · Cited by 382 · see citing works
ASVspoof 2021: accelerating progress in spoofed and deepfake speech detection
Yamagishi, J., Wang, X., Todisco, M., Sahidullah, M., Patino, J., Nautsch, A., Liu, X., Lee, K. A., Kinnunen, T., Evans, N., et al. (2021) · 2021 Edition of the Automatic Speaker Verification and Spoofing Countermeasures Challenge · doi:10.21437/asvspoof.2021-8 · Cited by 380 · see citing works
Continual learning of context-dependent processing in neural networks
Zeng, G., Chen, Y., Cui, B., & Yu, S. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-019-0080-x · Cited by 309 · see citing works
Does Audio Deepfake Detection Generalize?
Müller, N., Czempin, P., Diekmann, F., Froghyar, A., & Böttinger, K. (2022) · Interspeech 2022 · doi:10.21437/interspeech.2022-108 · Cited by 258 · see citing works
One-class learning towards synthetic voice spoofing detection
Zhang, Y., Jiang, F., & Duan, Z. (2021) · IEEE Signal Processing Letters · doi:10.1109/lsp.2021.3076358 · Cited by 241 · see citing works
Tak, H., Todisco, M., Wang, X., Jung, J.-W., Yamagishi, J., & Evans, N. (2022) · The Speaker and Language Recognition Workshop (Odyssey 2022) · doi:10.21437/odyssey.2022-16 · Cited by 237 · see citing works
Optimal filtering algorithms for fast learning in feedforward neural networks
Shah, S., Palmieri, F., & Datum, M. (1992) · Neural Networks · doi:10.1016/s0893-6080(05)80139-x · Cited by 172 · see citing works
ADD 2022: the first Audio Deep Synthesis Detection Challenge
Yi, J., Fu, R., Tao, J., Nie, S., Ma, H., Wang, C., Wang, T., Tian, Z., Bai, Y., Fan, C., et al. (2022) · ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) · doi:10.1109/icassp43922.2022.9746939 · Cited by 165 · see citing works
Self-citationThe Vicomtech Audio Deepfake Detection System Based on Wav2vec2 for the 2022 ADD ChallengePreprint
Martín-Doñas, J. M., & Álvarez, A. (2022) · IEEE International Conference on Acoustics, Speech, and Signal Processing · doi:10.48550/arxiv.2203.01573 · Cited by 155 · see citing works
Investigating Self-Supervised Front Ends for Speech Spoofing Countermeasures
Wang, X., & Yamagishi, J. (2022) · The Speaker and Language Recognition Workshop (Odyssey 2022) · doi:10.21437/odyssey.2022-14 · Cited by 138 · see citing works
The Vicomtech Audio Deepfake Detection System Based on Wav2vec2 for the 2022 ADD Challenge
Martin-Donas, J. M., & Alvarez, A. (2022) · ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) · doi:10.1109/icassp43922.2022.9747768 · Cited by 100 · see citing works
Fake Audio Detection Based On Unsupervised Pretraining Models
Lv, Z., Zhang, S., Tang, K., & Hu, P. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9747605 · Cited by 61 · see citing works
Continual Learning for Fake Audio Detection
Ma, H., Yi, J., Tao, J., Bai, Y., Tian, Z., & Wang, C. (2021) · Interspeech 2021 · doi:10.21437/interspeech.2021-794 · Cited by 39 · see citing works
Self-citationThe CLEAR Benchmark: Continual LEArning on Real-World ImageryPreprint
Lin, Z., Shi, J. D., Pathak, D., & Ramanan, D. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2201.06289 · Cited by 25 · see citing works
ADD 2023: the Second Audio Deepfake Detection Challenge
Yi, J., Tao, J., Fu, R., Yan, X., Wang, C., Wang, T., Zhang, C. Y., Zhang, X., Zhao, Y., Ren, Y., et al. (2023) · arXiv · doi:10.48550/arxiv.2305.13774 · Cited by 20 · see citing works
Self-citationProsody and Voice Factorization for Few-Shot Speaker Adaptation in the Challenge M2voc 2021
Wang, T., Fu, R., Yi, J., Tao, J., Wen, Z., Qiang, C., & Wang, S. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9414427 · Cited by 17 · see citing works
Self-citationInvestigating self-supervised front ends for speech spoofing countermeasuresPreprint
Wang, X., & Yamagishi, J. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2111.07725 · Cited by 10 · see citing works
Tak, H., Todisco, M., Wang, X., et al. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2202.12233 · Cited by 4 · see citing works
Flagged first, then by citations
Research works printed but not linked: 2
- [18]18. Bischof, H.; Brox, T.; and Frahm, J., eds., Computer Vision -ECCV 2020 -16th Eu- ropean Conference, Glasgow, UK, August 23-28, 2020, Pro- ceedings, Part II, volume 12347 of Lecture Notes in Com- puter Science, 524-540. Springer.
- [32]32. Zeng, G.; Chen, Y.; Cui, B.; and Yu, S. 2019. Contin- ual learning of context-dependent processing in neural net- works. Nature Machine Intelligence, 1(8): 364-372.
Cited by
Sorted by citations
Deepfake Media Forensics: Status and Future Challenges
Amerini, I., Barni, M., Battiato, S., Bestagini, P., Boato, G., Bruni, V., Caldelli, R., De Natale, F., De Nicola, R., Guarnera, L., et al. (2025) · Journal of Imaging · doi:10.3390/jimaging11030073 · Cited by 113 · see citing works
Multimodal Representation Learning by Alternating Unimodal Adaptation
Zhang, X., Yoon, J., Bansal, M., & Yao, H. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.02592 · Cited by 71 · see citing works
Self-citationAudio Deepfake Detection: What Has Been Achieved and What Lies Ahead
Zhang, B., Cui, H., Nguyen, V., & Whitty, M. (2025) · Sensors · doi:10.3390/s25071989 · Cited by 72 · see citing works
Ding, Y., Guo, B., Liu, Y., Jing, Y., Yin, M., Li, N., Wang, H., & Yu, Z. (2024) · Information Processing & Management · doi:10.1016/j.ipm.2024.103878 · Cited by 21 · see citing works
Wani, T. M., Qadri, S. A. A., Wani, F. A., & Amerini, I. (2024) · Foundations and Trends® in Privacy and Security · doi:10.1561/3300000048 · Cited by 16 · see citing works
A comprehensive survey with critical analysis for deepfake speech detection
Pham, L. D., Lam, P., Tran, D., Tang, H., Nguyen, T., Schindler, A., Skopik, F., Polonsky, A., & Vu, H. C. (2025) · Computer Science Review · doi:10.1016/j.cosrev.2025.100757 · Cited by 23 · see citing works
Advancements in AI-Generated Content Forensics: A Systematic Literature Review
Xu, Q., Mu, W., Li, J., Sun, T., & Jiang, X. (2025) · ACM Computing Surveys · doi:10.1145/3760526 · Cited by 16 · see citing works
Deepfake Audio Detection in Voice Authentication: A Spectral and CNN-Based Comprehensive Review
Salih, A. O. M., Emam, A. H. M., Ahmed, A. B. G. E., Khalifa, M., Suliman, A., & Babiker, N. B. M. (2025) · Engineering Technology & Applied Science Research · doi:10.48084/etasr.13400 · Cited by 8 · see citing works
Adaptive spectral bandpass multi-scale network for underwater acoustic target recognition
Qi, P., Yin, G., Zhang, L., Dong, Y., & Zhang, L. (2025) · Multimedia Systems · doi:10.1007/s00530-025-02067-y · Cited by 3 · see citing works
Wani, T. M., & Amerini, I. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-78305-0_14 · Cited by 3 · see citing works
Generalized Audio Deepfake Detection Using Frame-level Latent Information Entropy
Zhao, B., Kang, Z., He, Y., Qu, X., Peng, J., Xiao, J., & Wang, J. (2025) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme59968.2025.11210096 · Cited by 3 · see citing works
Freeze and Learn: Continual Learning with Selective Freezing for Speech Deepfake Detection
Salvi, D., Negroni, V., Bondi, L., Bestagini, P., & Tubaro, S. (2025) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49660.2025.10889357 · Cited by 4 · see citing works
A Two-Stage LoRA Strategy for Expanding Language Capabilities in Multilingual ASR Models
Kwok, C. Y., Liu, H., Yip, J. Q., Li, S., & Chng, E. S. (2025) · IEEE Transactions on Audio Speech and Language Processing · doi:10.1109/taslpro.2025.3578752 · Cited by 3 · see citing works
Lee, H.-J., & Buu, S.-J. (2024) · IEEE Access · doi:10.1109/access.2024.3459954 · Cited by 3 · see citing works
Li, Z., Teng, Z., Zhang, B., & Jianping Fan, B. (2024) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3664647.3680737 · Cited by 2 · see citing works
Artificial Intelligence and Communication: The Threat of Deepfake Images
Domínguez, L. M., Andión, J. L. Z., & Kolankowska, M. (2025) · Research series on responsible enterprise ecosystems · doi:10.1007/978-3-031-80411-3_7 · Cited by 1 · see citing works
Unmasking Synthetic Speech: A Novel Deep Learning Model for Deepfake Audio Detection
R, R. P. N., Priyadharshini, J. M. H., Arun Manicka Raja, M., & Shakthi Priya, S. (2025) · International Conference on Research Methodologies in Knowledge Management, Artificial Intelligence and Telecommunication Engineering (RMKMATE) · doi:10.1109/rmkmate64874.2025.11042838 · Cited by 1 · see citing works
ASTDT: an Interpretable Adaptive Spectro-Temporal Diffusion Transformer for audio deepfake detection
Wani, T. M., Qadri, S. A. A., Ashraf, A., & Amerini, I. (2025) · EURASIP Journal on Information Security · doi:10.1186/s13635-025-00217-3 · Cited by 2 · see citing works
Zero-Shot to Zero-Lies: Detecting Bengali Deepfake Audio Through Transfer Learning
Samu, M. S. S., Islam, M. R., Hossain, M. Z., Bhuiyan, M. K., & Zaman, F. U. (2025) · International Conference on Computer and Information Technology (ICCIT) · doi:10.1109/iccit68739.2025.11491615 · Cited by 2 · see citing works
Gazis, A., Chamalidou, E., Ntaoulas, N., & Vavouras, T. (2026) · Academia AI and Applications · doi:10.20935/acadai8365 · Cited by 1 · see citing works
Perturbed Public Voices (P 2 V): A Dataset for Robust Audio Deepfake Detection
Gao, C., Postiglione, M., Gortner, I., Kraus, S., & Subrahmanian, V. S. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770855.3817542 · Cited by 1 · see citing works
SHDA: Sinkhorn Domain Attention for Cross-Domain Audio Anti-Spoofing
Zhang, R., Wei, J., Lu, X., Zhang, L., Jin, D., Xu, J., & Lu, W. (2025) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2025.3576576 · Cited by 1 · see citing works
Information Bottleneck Based Data Correction in Continual Learning
Chen, S., Zhang, M., Zhang, J., & Huang, K. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-73021-4_16
Low-Resource Language Adaptation Using Two-Stage Regularization for Multilingual ASR
Kwok, C. Y., Yip, J. Q., & Chng, E. S. (2025) · International Journal of Asian Language Processing · doi:10.1142/s2717554525500031
Xiao, Y., & Vu, N. T. (2025) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49660.2025.10888114
Ghallab, H., Nasr, M. M., & Fahmy, H. (2025) · International Journal of Advanced Computer Science and Applications · doi:10.14569/ijacsa.2025.0160414
Enhancing Deepfake Detection: A Study Using WavLM and Advanced RawBoost Augmentation Techniques
Tri, N., Hoang, L. N., Viet, P. T., & Trung, K. P. (2024) · RIVF International Conference on Computing and Communication Technologies (RIVF) · doi:10.1109/rivf64335.2024.11009068
Continual Audio Deepfake Detection via Universal Adversarial Perturbation
Li, W., Li, L., & Hong, Q. (2025) · Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC) · doi:10.1109/apsipaasc65261.2025.11249335
What to Remember: Self-Adaptive Continual Learning for Audio Deepfake Detection
Association for Artificial Intelligence 2024, Tao, J., Wang, C., Yi, J., Zeng, S., Zhang, X., & Zhang, C. Y. (2024) · Underline Science Inc. · doi:10.48448/7856-1459
Self-citationAFT: An Exemplar-Free Class Incremental Learning Method for Environmental Sound Classification
Chen, X., Chen, X., Weng, Z., & Xiao, Y. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11460616
Zhang, Y.-X., Zhou, Z., Luo, H., Liu, W., & Li, M. (2026) · Applied Intelligence · doi:10.1007/s10489-026-07295-x
Jajoo, P., Taksalee, P., Singh, S., Hossain, A. M., & Quader, F. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-34132-7_36
- Authors
- Zhang, X., Yi, J., Wang, C., Zhang, C. Y., Zeng, S., & Tao, J.
- Affiliations
- Chinese Academy of Sciences (CN)
- Tsinghua University (CN)
- Venue
- Proceedings of the AAAI Conference on Artificial Intelligence
- Published
- 2024-03-24
- Volume
- 38
- Issue
- 17
- Pages
- 19569-19577
- ISSN
- 2159-5399 (ISSN-L)
- 2374-3468
- OpenAlex
- W4393160839
- Cited by
- 32
- References
- 34
- Citations by year
- 2026: 5 · 2025: 18 · 2024: 9
- FWCI
- 9.24
- Type
- conference paper
- Language
- en
- Open access
- diamond
- doi.org
- Locations
- Proceedings of the AAAI Conference on Artificial Intelligence (publishedVersion)
- arXiv (Cornell University) (submittedVersion, cc-by)
- AAAI Publications (The Association for the Advancement of Artificial Intelligence (AAAI)) (submittedVersion)
- arXiv (Cornell University) (cc-by)
- Funders
- National Natural Science Foundation of China (U21B2010, 61971419, 62306316, 61831022, 62322120, 62206278, 62101553, 62006223)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Speech Recognition and Synthesis
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Physical Sciences › Computer Science › Signal Processing › Music and Audio Processing
- Keywords
- audio deepfake detection, continual learning, self-adaptive learning, knowledge forgetting, speech synthesis, voice conversion
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
