Audio and Voice Deepfake Detection: a Survey on Cross-Dataset Generalization Bottlenecks
Resumo
A geração de fala sintética tornou-se acessível o suficiente para viabilizar fraudes e desinformação, intensificando a necessidade de detectores robustos de deepfake de áudio. Este artigo apresenta uma revisão conduzida segundo o protocolo PRISMA 2020 sobre as limitações de generalização cross-dataset em detectores de deepfake de áudio e voz, com 37 estudos incluídos. Os estudos que decompõem a queda de desempenho atribuem-na sobretudo à diferença de distribuição entre corpora, e não à dificuldade intrínseca do alvo: a Taxa de Erro de Igualdade (in-domain) próxima de 1% degrada para 27%–34% em conjuntos out-of-distribution. Os front-ends auto-supervisionados e a diversificação de dados são as estratégias mais consistentemente associadas a ganhos entre os estudos incluídos, embora nenhuma resolva o gargalo isoladamente. Mostra-se ainda que a comparabilidade é comprometida por inconsistências de protocolos e métricas, e propõe-se uma agenda para protocolos cross-corpus padronizados.
Referências
Badugu, V. K., Veesa, S., and Singh, M. (2025). The Kolmogorov-Arnold networks: A new foundation paradigm for hierarchical function learning in ASV anti-spoofing research. In 2025 IEEE International Conference on Communication, Networks and Satellite (COMNETSAT), pages 359–366. IEEE.
Conti, E., Salvi, D., Borrelli, C., Hosler, B., Bestagini, P., Antonacci, F., Sarti, A., Stamm, M. C., and Tubaro, S. (2022). Deepfake speech detection through emotion recognition: A semantic approach. In ICASSP 2022 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 8962–8966. IEEE.
Dutta, B., Ranjan, R., Sathvik, S., Vatsa, M., and Singh, R. (2025). Can quantized audio language models perform zero-shot spoofing detection? In Interspeech 2025, pages 4578–4582. ISCA.
Ge, W., Wang, X., Yamagishi, J., Todisco, M., and Evans, N. (2024). Spoofing attack augmentation: Can differently-trained attack models improve generalisation? In ICASSP 2024 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 12531–12535. IEEE.
Gong, L. Y. and Li, X. J. (2025). Deepfake voice detection: An approach using end-to-end transformer with acoustic feature fusion by cross-attention. Electronics, 14(10):2040.
Huang, S.-F., Kuo, H.-C., Chen, Z., Yang, X., Yang, C.-H. H., Tsao, Y., Wang, Y.-C. F., Lee, H.-Y., and Fu, S.-W. (2024). Detecting the undetectable: Assessing the efficacy of current spoof detection methods against seamless speech edits. In 2024 IEEE Spoken Language Technology Workshop (SLT), pages 652–659. IEEE.
Jung, J.-w., Heo, H.-S., Tak, H., Shim, H.-j., Chung, J. S., Lee, B.-J., Yu, H.-J., and Evans, N. (2022). AASIST: Audio anti-spoofing using integrated spectro-temporal graph attention networks. In ICASSP 2022 – IEEE International Conference on Acoustics, Speech and Signal Processing. IEEE.
Kang, J. Y., Yoon, J. W., Kim, S., Han, M. H., and Kim, N. S. (2025). FADEL: Uncertainty-aware fake audio detection with evidential deep learning. In ICASSP 2025 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 1–5. IEEE.
Kawa, P., Plata, M., Czuba, M., Szymański, P., and Syga, P. (2023). Improved deepfake detection using Whisper features. In Interspeech 2023, pages 4009–4013. ISCA.
Khan, A., Malik, K. M., and Nawaz, S. (2024). Frame-to-utterance convergence: A spectra-temporal approach for unified spoofing detection. In ICASSP 2024 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 10761–10765. IEEE.
Khan, A., Malik, K. M., Ryan, J., and Saravanan, M. (2023). Battling voice spoofing: A review, comparative analysis, and generalizability evaluation of state-of-the-art voice spoofing counter measures. Artificial Intelligence Review, 56(S1):513–566.
Kitchenham, B. and Charters, S. (2007). Guidelines for performing systematic literature reviews in software engineering. Technical Report EBSE 2007-001, Keele University and Durham University.
Li, X., Li, N., Weng, C., Liu, X., Su, D., Yu, D., and Meng, H. (2021). Replay and synthetic speech detection with Res2Net architecture. In ICASSP 2021 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 6354–6358. IEEE.
Liu, X., Wang, X., and Yamagishi, J. (2024). A preliminary case study on long-form in-the-wild audio spoofing detection. In 2024 International Conference of the Biometrics Special Interest Group (BIOSIG), pages 1–6. IEEE.
Luong, H.-T., Rimon, I., Permuter, H., Lee, K. A., and Chng, E. S. (2025). Robust localization of partially fake speech: Metrics and out-of-domain evaluation. In 2025 Asia Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC), pages 2205–2210. IEEE.
Müller, N., Czempin, P., Diekmann, F., Froghyar, A., and Böttinger, K. (2022). Does audio deepfake detection generalize? In Interspeech 2022, pages 2783–2787. ISCA.
Müller, N. M., Evans, N., Tak, H., Sperl, P., and Böttinger, K. (2024). Harder or different? understanding generalization of audio deepfake detection. In Interspeech 2024, pages 2705–2709. ISCA.
Müller, N. M., Sperl, P., and Böttinger, K. (2023). Complex-valued neural networks for voice anti-spoofing. In Interspeech 2023, pages 3814–3818. ISCA.
Page, M. J., McKenzie, J. E., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., et al. (2021a). The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ, 372:n71.
Page, M. J., Moher, D., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., et al. (2021b). PRISMA 2020 explanation and elaboration: Updated guidance and exemplars for reporting systematic reviews. BMJ, 372:n160.
Park, S. and Kwak, I.-Y. (2025). Impact of neural-vocoder-generated spoofed speech on spoofing-detection performance. Korean Journal of Applied Statistics, 38(5):679–692.
Ranjan, R., Vatsa, M., and Singh, R. (2022). STATNet: Spectral and temporal features based multi-task network for audio spoofing detection. In 2022 IEEE International Joint Conference on Biometrics (IJCB), pages 1–9. IEEE.
Ranjan, R., Vatsa, M., and Singh, R. (2023). Uncovering the deceptions: An analysis on audio spoofing detection and future prospects. In Proceedings of the 32nd International Joint Conference on Artificial Intelligence (IJCAI), Survey Track, pages 6750–6758.
Sahidullah, M., Delgado, H., Todisco, M., Nautsch, A., Wang, X., Kinnunen, T., Evans, N., Yamagishi, J., and Lee, K.-A. (2023). Introduction to voice presentation attack detection and recent advances. In Handbook of Biometric Anti-Spoofing, Advances in Computer Vision and Pattern Recognition, pages 339–385. Springer.
Sali, Y., Erden, M., and Haznedaroğlu, A. (2023). Investigation of approaches in the literature and potential problems for audio spoofing detection. In 2023 31st Signal Processing and Communications Applications Conference (SIU), pages 1–4. IEEE.
Salvi, D., Bestagini, P., and Tubaro, S. (2023). Towards frequency band explainability in synthetic speech detection. In 2023 31st European Signal Processing Conference (EUSIPCO), pages 620–624. IEEE.
Shih, T.-H., Yeh, C.-Y., and Chen, M.-S. (2024). Does audio deepfake detection rely on artifacts? In ICASSP 2024 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 12446–12450. IEEE.
Shim, H.-j., Jung, J.-w., and Kinnunen, T. (2023). Multi-dataset co-training with sharpness-aware optimization for audio anti-spoofing. In Interspeech 2023, pages 3804–3808. ISCA.
Singh Yadav, A. K., Bhagtani, K., Bestagini, P., and Delp, E. J. (2024a). Improving fairness in synthetic speech detectors. In 2024 58th Asilomar Conference on Signals, Systems, and Computers, pages 362–366. IEEE.
Singh Yadav, A. K., Bhagtani, K., Salvi, D., Bestagini, P., and Delp, E. J. (2024b). FairSSD: Understanding bias in synthetic speech detectors. In 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 4418–4428. IEEE.
Singh Yadav, A. K., Xiang, Z., Bhagtani, K., Bestagini, P., Tubaro, S., and Delp, E. J. (2023). PS3DT: Synthetic speech detection using patched spectrogram transformer. In 2023 International Conference on Machine Learning and Applications (ICMLA), pages 496–503. IEEE.
Tak, H., Todisco, M., Wang, X., Jung, J.-w., Yamagishi, J., and Evans, N. (2022). Automatic speaker verification spoofing and deepfake detection using wav2vec 2.0 and data augmentation. In The Speaker and Language Recognition Workshop (Odyssey 2022).
Wang, H., Dinkel, H., Wang, S., Qian, Y., and Yu, K. (2019). Cross-domain replay spoofing attack detection using domain adversarial training. In Interspeech 2019, pages 2938–2942. ISCA.
Wang, X. and Yamagishi, J. (2023). Investigating active-learning-based training data selection for speech spoofing countermeasure. In 2022 IEEE Spoken Language Technology Workshop (SLT), pages 585–592. IEEE.
Wang, X. and Yamagishi, J. (2024). Can large-scale vocoded spoofed data improve speech spoofing countermeasure with a self-supervised front end? In ICASSP 2024 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 10311–10315. IEEE.
Wang, Y., Lin, X., Qiu, Y., and Tan, G. (2026). Leveraging Gaussian attention for adapter-based fine-tuning of WavLM in audio deepfake detection. International Journal of Intelligent Systems, 2026(1):6786731.
Wu, H., Kuo, H.-C., Zheng, N., Hung, K.-H., Lee, H.-Y., Tsao, Y., Wang, H.-M., and Meng, H. (2022). Partially fake audio detection by self-attention-based fake span discovery. In ICASSP 2022 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 9236–9240. IEEE.
Xie, Y., Lu, Y., Fu, R., Wen, Z., Wang, Z., Tao, J., Qi, X., Wang, X., Liu, Y., Cheng, H., Ye, L., and Sun, Y. (2025). The Codecfake dataset and countermeasures for the universally detection of deepfake audio. IEEE Transactions on Audio, Speech and Language Processing, 33:386–400.
Yamagishi, J., Wang, X., Todisco, M., Sahidullah, M., Patino, J., et al. (2021). ASVspoof 2021: Accelerating progress in spoofed and deepfake speech detection. In ASVspoof 2021 Workshop.
Yi, J., Wang, C., Tao, J., Zhang, X., Zhang, C. Y., and Zhao, Y. (2023). Audio deepfake detection: A survey. arXiv preprint arXiv:2308.14970.
Zang, Y., Zhang, Y., Heydari, M., and Duan, Z. (2024). SingFake: Singing voice deepfake detection. In ICASSP 2024 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 12156–12160. IEEE.
Zeng, C., Wang, X., Miao, X., Cooper, E., and Yamagishi, J. (2023). Improving generalization ability of countermeasures for new mismatch scenario by combining multiple advanced regularization terms. In Interspeech 2023, pages 1998–2002. ISCA.
Zhang, X., Yi, J., Tao, J., Wang, C., and Zhang, C. Y. (2023). Do you remember? overcoming catastrophic forgetting for fake audio detection. In Proceedings of the 40th International Conference on Machine Learning (ICML), volume 202 of PMLR, pages 41819–41831.
Zhang, Y., Zhu, G., Jiang, F., and Duan, Z. (2021). An empirical study on channel effects for synthetic voice spoofing countermeasure systems. In Interspeech 2021, pages 4309–4313. ISCA.
Conti, E., Salvi, D., Borrelli, C., Hosler, B., Bestagini, P., Antonacci, F., Sarti, A., Stamm, M. C., and Tubaro, S. (2022). Deepfake speech detection through emotion recognition: A semantic approach. In ICASSP 2022 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 8962–8966. IEEE.
Dutta, B., Ranjan, R., Sathvik, S., Vatsa, M., and Singh, R. (2025). Can quantized audio language models perform zero-shot spoofing detection? In Interspeech 2025, pages 4578–4582. ISCA.
Ge, W., Wang, X., Yamagishi, J., Todisco, M., and Evans, N. (2024). Spoofing attack augmentation: Can differently-trained attack models improve generalisation? In ICASSP 2024 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 12531–12535. IEEE.
Gong, L. Y. and Li, X. J. (2025). Deepfake voice detection: An approach using end-to-end transformer with acoustic feature fusion by cross-attention. Electronics, 14(10):2040.
Huang, S.-F., Kuo, H.-C., Chen, Z., Yang, X., Yang, C.-H. H., Tsao, Y., Wang, Y.-C. F., Lee, H.-Y., and Fu, S.-W. (2024). Detecting the undetectable: Assessing the efficacy of current spoof detection methods against seamless speech edits. In 2024 IEEE Spoken Language Technology Workshop (SLT), pages 652–659. IEEE.
Jung, J.-w., Heo, H.-S., Tak, H., Shim, H.-j., Chung, J. S., Lee, B.-J., Yu, H.-J., and Evans, N. (2022). AASIST: Audio anti-spoofing using integrated spectro-temporal graph attention networks. In ICASSP 2022 – IEEE International Conference on Acoustics, Speech and Signal Processing. IEEE.
Kang, J. Y., Yoon, J. W., Kim, S., Han, M. H., and Kim, N. S. (2025). FADEL: Uncertainty-aware fake audio detection with evidential deep learning. In ICASSP 2025 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 1–5. IEEE.
Kawa, P., Plata, M., Czuba, M., Szymański, P., and Syga, P. (2023). Improved deepfake detection using Whisper features. In Interspeech 2023, pages 4009–4013. ISCA.
Khan, A., Malik, K. M., and Nawaz, S. (2024). Frame-to-utterance convergence: A spectra-temporal approach for unified spoofing detection. In ICASSP 2024 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 10761–10765. IEEE.
Khan, A., Malik, K. M., Ryan, J., and Saravanan, M. (2023). Battling voice spoofing: A review, comparative analysis, and generalizability evaluation of state-of-the-art voice spoofing counter measures. Artificial Intelligence Review, 56(S1):513–566.
Kitchenham, B. and Charters, S. (2007). Guidelines for performing systematic literature reviews in software engineering. Technical Report EBSE 2007-001, Keele University and Durham University.
Li, X., Li, N., Weng, C., Liu, X., Su, D., Yu, D., and Meng, H. (2021). Replay and synthetic speech detection with Res2Net architecture. In ICASSP 2021 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 6354–6358. IEEE.
Liu, X., Wang, X., and Yamagishi, J. (2024). A preliminary case study on long-form in-the-wild audio spoofing detection. In 2024 International Conference of the Biometrics Special Interest Group (BIOSIG), pages 1–6. IEEE.
Luong, H.-T., Rimon, I., Permuter, H., Lee, K. A., and Chng, E. S. (2025). Robust localization of partially fake speech: Metrics and out-of-domain evaluation. In 2025 Asia Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC), pages 2205–2210. IEEE.
Müller, N., Czempin, P., Diekmann, F., Froghyar, A., and Böttinger, K. (2022). Does audio deepfake detection generalize? In Interspeech 2022, pages 2783–2787. ISCA.
Müller, N. M., Evans, N., Tak, H., Sperl, P., and Böttinger, K. (2024). Harder or different? understanding generalization of audio deepfake detection. In Interspeech 2024, pages 2705–2709. ISCA.
Müller, N. M., Sperl, P., and Böttinger, K. (2023). Complex-valued neural networks for voice anti-spoofing. In Interspeech 2023, pages 3814–3818. ISCA.
Page, M. J., McKenzie, J. E., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., et al. (2021a). The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ, 372:n71.
Page, M. J., Moher, D., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., et al. (2021b). PRISMA 2020 explanation and elaboration: Updated guidance and exemplars for reporting systematic reviews. BMJ, 372:n160.
Park, S. and Kwak, I.-Y. (2025). Impact of neural-vocoder-generated spoofed speech on spoofing-detection performance. Korean Journal of Applied Statistics, 38(5):679–692.
Ranjan, R., Vatsa, M., and Singh, R. (2022). STATNet: Spectral and temporal features based multi-task network for audio spoofing detection. In 2022 IEEE International Joint Conference on Biometrics (IJCB), pages 1–9. IEEE.
Ranjan, R., Vatsa, M., and Singh, R. (2023). Uncovering the deceptions: An analysis on audio spoofing detection and future prospects. In Proceedings of the 32nd International Joint Conference on Artificial Intelligence (IJCAI), Survey Track, pages 6750–6758.
Sahidullah, M., Delgado, H., Todisco, M., Nautsch, A., Wang, X., Kinnunen, T., Evans, N., Yamagishi, J., and Lee, K.-A. (2023). Introduction to voice presentation attack detection and recent advances. In Handbook of Biometric Anti-Spoofing, Advances in Computer Vision and Pattern Recognition, pages 339–385. Springer.
Sali, Y., Erden, M., and Haznedaroğlu, A. (2023). Investigation of approaches in the literature and potential problems for audio spoofing detection. In 2023 31st Signal Processing and Communications Applications Conference (SIU), pages 1–4. IEEE.
Salvi, D., Bestagini, P., and Tubaro, S. (2023). Towards frequency band explainability in synthetic speech detection. In 2023 31st European Signal Processing Conference (EUSIPCO), pages 620–624. IEEE.
Shih, T.-H., Yeh, C.-Y., and Chen, M.-S. (2024). Does audio deepfake detection rely on artifacts? In ICASSP 2024 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 12446–12450. IEEE.
Shim, H.-j., Jung, J.-w., and Kinnunen, T. (2023). Multi-dataset co-training with sharpness-aware optimization for audio anti-spoofing. In Interspeech 2023, pages 3804–3808. ISCA.
Singh Yadav, A. K., Bhagtani, K., Bestagini, P., and Delp, E. J. (2024a). Improving fairness in synthetic speech detectors. In 2024 58th Asilomar Conference on Signals, Systems, and Computers, pages 362–366. IEEE.
Singh Yadav, A. K., Bhagtani, K., Salvi, D., Bestagini, P., and Delp, E. J. (2024b). FairSSD: Understanding bias in synthetic speech detectors. In 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 4418–4428. IEEE.
Singh Yadav, A. K., Xiang, Z., Bhagtani, K., Bestagini, P., Tubaro, S., and Delp, E. J. (2023). PS3DT: Synthetic speech detection using patched spectrogram transformer. In 2023 International Conference on Machine Learning and Applications (ICMLA), pages 496–503. IEEE.
Tak, H., Todisco, M., Wang, X., Jung, J.-w., Yamagishi, J., and Evans, N. (2022). Automatic speaker verification spoofing and deepfake detection using wav2vec 2.0 and data augmentation. In The Speaker and Language Recognition Workshop (Odyssey 2022).
Wang, H., Dinkel, H., Wang, S., Qian, Y., and Yu, K. (2019). Cross-domain replay spoofing attack detection using domain adversarial training. In Interspeech 2019, pages 2938–2942. ISCA.
Wang, X. and Yamagishi, J. (2023). Investigating active-learning-based training data selection for speech spoofing countermeasure. In 2022 IEEE Spoken Language Technology Workshop (SLT), pages 585–592. IEEE.
Wang, X. and Yamagishi, J. (2024). Can large-scale vocoded spoofed data improve speech spoofing countermeasure with a self-supervised front end? In ICASSP 2024 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 10311–10315. IEEE.
Wang, Y., Lin, X., Qiu, Y., and Tan, G. (2026). Leveraging Gaussian attention for adapter-based fine-tuning of WavLM in audio deepfake detection. International Journal of Intelligent Systems, 2026(1):6786731.
Wu, H., Kuo, H.-C., Zheng, N., Hung, K.-H., Lee, H.-Y., Tsao, Y., Wang, H.-M., and Meng, H. (2022). Partially fake audio detection by self-attention-based fake span discovery. In ICASSP 2022 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 9236–9240. IEEE.
Xie, Y., Lu, Y., Fu, R., Wen, Z., Wang, Z., Tao, J., Qi, X., Wang, X., Liu, Y., Cheng, H., Ye, L., and Sun, Y. (2025). The Codecfake dataset and countermeasures for the universally detection of deepfake audio. IEEE Transactions on Audio, Speech and Language Processing, 33:386–400.
Yamagishi, J., Wang, X., Todisco, M., Sahidullah, M., Patino, J., et al. (2021). ASVspoof 2021: Accelerating progress in spoofed and deepfake speech detection. In ASVspoof 2021 Workshop.
Yi, J., Wang, C., Tao, J., Zhang, X., Zhang, C. Y., and Zhao, Y. (2023). Audio deepfake detection: A survey. arXiv preprint arXiv:2308.14970.
Zang, Y., Zhang, Y., Heydari, M., and Duan, Z. (2024). SingFake: Singing voice deepfake detection. In ICASSP 2024 – IEEE International Conference on Acoustics, Speech and Signal Processing, pages 12156–12160. IEEE.
Zeng, C., Wang, X., Miao, X., Cooper, E., and Yamagishi, J. (2023). Improving generalization ability of countermeasures for new mismatch scenario by combining multiple advanced regularization terms. In Interspeech 2023, pages 1998–2002. ISCA.
Zhang, X., Yi, J., Tao, J., Wang, C., and Zhang, C. Y. (2023). Do you remember? overcoming catastrophic forgetting for fake audio detection. In Proceedings of the 40th International Conference on Machine Learning (ICML), volume 202 of PMLR, pages 41819–41831.
Zhang, Y., Zhu, G., Jiang, F., and Duan, Z. (2021). An empirical study on channel effects for synthetic voice spoofing countermeasure systems. In Interspeech 2021, pages 4309–4313. ISCA.
Publicado
19/10/2026
Como Citar
CASTRO, João Victor F. de; ROCHA, Pedro Luiz Seabra; LIMA, Jean P. de O.; AZAMBUJA, Antônio João G. de; OLIVEIRA, Leonardo B.; SOARES, Anderson da Silva.
Audio and Voice Deepfake Detection: a Survey on Cross-Dataset Generalization Bottlenecks. In: ENCONTRO NACIONAL DE INTELIGÊNCIA ARTIFICIAL E COMPUTACIONAL (ENIAC), 23. , 2026, Cuiabá/MT.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 3346-3357.
ISSN 2763-9061.
DOI: https://doi.org/10.5753/eniac.2026.31346.
