I-vectorlerin Maskeleme Yoluyla Dayanıklılığının Arttırılması: Sentetik Konuşma Tespitinde Bir Vaka Çalışması

Year 2024, Volume: 29 Issue: 1, 191 - 204, 22.04.2024

Barış Aydın Gökay Dişken

https://doi.org/10.17482/uumfd.1311113

Abstract

Konuşmacı tanıma sistemleri için güvenlik hayati önem taşımaktadır. Geçtiğimiz yıllarda, sahte konuşma saldırılarının bu sistemleri kandırabildiği ortaya konmuştur. Bu durumu önlemek amacı ile sahte konuşma tespit sistemleri geliştirilmiştir. Bu tür sistemler bazı durumlarda oldukça yüksek performans sergilese de, gürültü altında performansları kötüleşmektedir. Geleneksel konuşma iyileştirme yöntemleri performansı artırmak bir yana, daha da kötüleştirmektedir. Bu çalışmada, konvolüsyonel sinir ağı yapısı kullanılarak elde edilen maskenin gürültü etkisini azaltmaktaki performansı incelenmiştir. Maske, spektrogramın gürültülü bölgelerini bastırmakta ve bu spektrogramdan elde edilen i-vectorleri gürbüz hale getirmekte kullanılmıştır. ASVspoof 2015 veri tabanı ve üç farklı gürültü tipi ile gerçekleştirilen testlerde önerilen sistemin geleneksel sistemlerden daha üstün olduğu gösterilmiştir. Ancak eğitim aşamasında karşılaşılmayan gürültü tiplerinde performans kaybı olmaktadır.

Keywords

derin öğrenme, evrişimli sinir ağı, sahte konuşma tanıma, konuşmacı tanıma, gürbüz öznitelikler

Project Number

121E057

INCREASING ROBUSTNESS OF I-VECTORS VIA MASKING: A CASE STUDY IN SYNTHETIC SPEECH DETECTION

Abstract

Ensuring security in speaker recognition systems is crucial. In the past years, it has been demonstrated that spoofing attacks can fool these systems. In order to deal with this issue, spoof speech detection systems have been developed. While these systems have served with a good performance, their effectiveness tends to degrade under noise. Traditional speech enhancement methods are not efficient for improving performance, they even make it worse. In this research paper, performance of the noise mask obtained via a convolutional neural network structure for reducing the noise effects was investigated. The mask is used to suppress noisy regions of spectrograms in order to extract robust i-vectors. The proposed system is tested on the ASVspoof 2015 database with three different noise types and accomplished superior performance compared to the traditional systems. However, there is a loss of performance in noise types that are not encountered during training phase.

Keywords

deep learning, convolutional neural network, spoof detection, speaker recognition, robust features

Supporting Institution

TÜBİTAK

Project Number

121E057

Thanks

This work was supported by TÜBİTAK (Project No: 121E057).

References

• 1. Alegre, F., Amehraye, A. and Evans, N. (2013) A one-class classification approach to generalized speaker verification spoofing countermeasures using local binary patterns, PInt. Conf. on Biometrics: Theory, Applications and Systems (BTAS), IEEE, Washington DC, USA. doi: 10.1109/BTAS.2013.6712706
• 2. ASVspoof, (2014). ASVspoof 2015: Automatic speaker verification spoofing and countermeasures challenge evaluation plan. Available: https://www.asvspoof.org/asvSpoof.pdf Accessed: Dec 19, 2014
• 3. Benhafid, Z., Selouani, S. A., Yakoub, M. S., Amrouche, A. (2021) LARIHS ASSERT reassessment for logical access ASVspoof 2021 challenge. Proceedings of the 2021 Edition of the Automatic Speaker Verification and Spoofing Countermeasures Challenge, Online, 94-99. doi: 10.21437/ASVSPOOF.2021-15
• 4. Dean, D., Kanagasundaram, A., Ghaemmaghami, H., Hafizur, M., Sridharan, S. (2015) The QUT-NOISE-SRE protocol for the evaluation of noisy speaker recognition, Interspeech 2015, International Speech and Communication Association, Dresden. doi: 10.21437/Interspeech.2015-685
• 5. Dehak, N., Kenny, P. J., Dehak, R., Dumouchel, P., Ouellet, P. (2011) Front-End Factor Analysis for Speaker Verification, IEEE/ACM Transactions on Audio, Speech, and Language Processing, 19(4), 788-798. doi: 10.1109/TASL.2010.2064307
• 6. Delgado, H., Todisco, M., Sahidullah, M., Evans, N., Kinnunen, T., Lee, K. A., Yamagishi, J. (2018) ASVspoof 2017 Version 2.0: meta-data analysis and baseline enhancements, Odyssey 2018 - The Speaker and Language Recognition Workshop, ASVSpoof, Odyssey, 296-303. doi: 10.21437/Odyssey.2018-42
• 7. Dipjyoti, P., Monisankha, P., Goutam, S., (2015) Novel speech features for improved detection of spoofing attacks, 2015 Annual IEEE India Conference (INDICON), New Delhi, India, pp. 1-6, doi: 10.1109/INDICON.2015.7443805.
• 8. Dipjyoti, P., Monisankha, P., Goutam, S., (2017) Spectral features for synthetic speech detection. IEEE journal of selected topics in signal processing, 11.4: 605-617. doi: 10.1109/JSTSP.2017.2684705
• 9. Dişken, G. (2023) Differential convolutional network for noise mask estimation. Applied Acoustics, 211, 109568. doi: 10.1016/j.apacoust.2023.109568
• 10. Dutoit, T., Holzapfel, A., Jottrand, M., Moinet, A., Perez, J., Stylianou, Y. (2007) Towards a voice conversion system based on frame selection, Proc. IEEE Int. Conf. on Acoustics, Speech, and Signal Processing (ICASSP),. IEEE, Honolulu, USA,4, 513-516. doi: 10.1109/ICASSP.2007.366962
• 11. Evans, N., Yamagishi, J., and Kinnunen, T. (2013) Spoofing and countermeasures for speaker verification: a need for standard corpora, protocols, and metrics, IEEE Signal Processing Society Speech and Language Technical Committee Newsletter..
• 12. Evans, N., Kinnunen, T., and Yamagishi, J. (2013) Spoofing and countermeasures for automatic speaker verification, Interspeech 2013, ISCA, Lyon, France, 925-929. doi: 10.21437/Interspeech.2013- 288.
• 13. Find Biometrics (2018). Voicevault Biometrics to Protect Payments. Available: https://findbiometrics.com/voicevault-biometrics-toprotect-payments-25131/ (Accessed: Jun. 13, 2018)
• 14. Find Biometrics (2018). Morpho and Agnitio Partner, Bring Voice Biometrics to Criminal ID. Available: https:// findbiometrics.com/morpho-and-agnitio-partner-bring-voice-biometricsto-criminal-id-21261/ (Accessed: Jun. 13, 2018)
• 15. Gomez-Alanis, A., Peinado, A. M., Gonzalez, J. A., and Gomez, A. M. (2018) A Deep Identity Representation for Noise Robust Spoofing Detection, Interspeech 2018, International Speech and Communication Association, Haydarabad, 676-680. doi: 10.21437/Interspeech.2018-1909
• 16. Gomez-Alanis, A., Peinado, A. M., Gonzalez, J. A., and Gomez, A. M. (2019) A Gated Recurrent Convolutional Neural Network for Robust Spoofing Detection, IEEE/ACM Transactions on Audio, Speech, And Language Processing, 27(12), 1985-1999. Doi: 10.1109/TASLP.2019.2937413
• 17. Hanilçi, C. (2018) Data selection for i-vector based automatic speaker verification anti-spoofing, Digital Signal Processing, 72, 171-180. doi: 10.1016/j.dsp.2017.10.010 (Article)
• 18. Hanilçi, C., Kinnunen, T., Sahidullaha, M., Sizova, A. (2016) Spoofing detection goes noisy: An analysis of synthetic speech detection in the presence of additive noise, Speech Communication, 85, 83-97. doi: 10.1016/j.specom.2016.10.002
• 19. Hassan, F., Javed, A. (2021) "Voice Spoofing Countermeasure for Synthetic Speech Detection," 2021 International Conference on Artificial Intelligence (ICAI), Islamabad, Pakistan, 2021, pp. 209-212, doi: 10.1109/ICAI52203.2021.9445238.
• 20. HSBC (2017). HSBC Voice ID Making Telephone Banking Safer Than Ever. Available: https://www.hsbc.co.uk/1/2/voice-id (Accessed: Dec. 29, 2017)
• 21. Jung, J., Heo, H., Tak, H., Shim, H., Chung, J. S., Lee, B. J., Yu, H. J., & Evans, N. (2022) AASIST: Audio Anti-Spoofing using Integrated Spectro-Temporal Graph Attention Networks, ICASSP 2022, IEEE, Lyon, France. doi: 10.21437/Interspeech.2013-288
• 22. Nugroho, K., Winarno, E., (2022) Spoofing Detection of Fake Speech Using Deep Neural Network Algorithm, 2022 International Seminar on Application for Technology of Information and Communication (iSemantic), Semarang, Indonesia, pp. 56-60. doi: 10.1109/iSemantic55962.2022.9920401.
• 23. Sizov, A., Khoury, E., Kinnunen, T ., Wu, Z. and Marcel, S. (2015) Joint speaker verification and anti-spoofing in the i-vector space, IEEE Transactions on Information Forensics and Security, IEEE Transactions on Information Forensics and Security, 10(4), 821-832. doi: 10.1109/TIFS.2015.2407362
• 24. Xiao, X., Tian, X., Du, S., Xu, H., Chng, E., Li, H. (2015). Spoofing speech detection using high dimensional magnitude and phase features: the NTU approach for ASVspoof 2015 challenge. In Interspeech (pp. 2052-2056). doi:10.21437/Interspeech.2015-465
• 25. Varga, A., Steeneken, H. J. M. (1993) Assessment for automatic speech recognition: II. NOISEX-92: A database and an experiment to study the effect of additive noise on speech recognition systems, Speech Communication, 12(43), 247-251. doi: 10.1016/0167-6393(93)90095-3
• 26. Wang, D. L., Kjems, U., Perdersen, M. S., Boldt, J. B., Lunner, T. (2009) Speech intelligibility in background noise with ideal binary time-frequency masking, J. Acoustical Soc. America, 125(4), 2336– 2347. doi: 10.1121/1.3083233
• 27. Wang, X., Yamagishi, J. (2021) A Comparative Study on Recent Neural Spoofing Countermeasures for Synthetic Speech Detection., Interspeech 2021, ISCA, Brno, Czech Republic, 4259-4263. doi: 10.21437/Interspeech.2021-702
• 28. Wu, Z., Khodabakhsh, A., Demiroglu, C., Yamagishi, J., Saito, D., Toda, T., King, S. (2015) SAS: A speaker verification spoofing database containing diverse attacks, IEEE Int. Conf. on Acoustics, Speech, and Signal Processing (ICASSP), IEEE, South Brisbane, Queensland, Australia, 9(5), 4440-4444. doi: 10.1109/ICASSP.2015.7178810.
• 29. Wu, Z., Kinnunen, T., Evans, N., & Yamagishi, J. (2015). Automatic Speaker Verification Spoofing and Countermeasures Challenge (ASVspoof 2015) Database, University of Edinburgh. The Centre for Speech Technology Research (CSTR). https://doi.org/10.7488/ds/298.
• 30. Zhang, C., Yu, C., Hansen, J. H. L. (2017) An Investigation of Deep-Learning Frameworks for Speaker Verification Antispoofing, IEEE Journal of Selected Topics in Signal Processing, 11, 684-694, 2017.
• 31. Zhang, Y., Jiang, F., Duan, Z. (2020) One-Class Learning Towards Synthetic Voice Spoofing Detection, in IEEE Signal Processing Letters, vol. 28, pp. 937-941, doi: 10.1109/LSP.2021.3076358.