Yüksek Çözünürlüklü Spektrogramlar ve XFeat Derin Öznitelikleri ile Ses Kopyala-Taşı Sahteciliği Tespiti

Arda Üstübioğlu

doi:10.31466/kfbd.1852460

TR EN

Yüksek Çözünürlüklü Spektrogramlar ve XFeat Derin Öznitelikleri ile Ses Kopyala-Taşı Sahteciliği Tespiti

Abstract

Ses kopyala–taşı sahteciliği aynı ses kaydı içinden alınan bir segmentin tekrar aynı kayda farklı bir zaman aralığında yapıştırılmasıyla gerçekleştirilmektedir. Sahteciliğin aynı kayıt içerisinde yapılması nedeniyle, kaynak/hedef segment aynı kanal ve arka plan karakteristiğine sahip olması sahtecilik tespitini oldukça güçleştirmektedir. Bu amaçla önerilen çalışmada, sinyal yüksek çözünürlüklü spektrogram görüntülerine dönüştürülerek XFeat tabanlı derin yerel tanımlayıcılarla tekrarlayan zaman–frekans örüntüleri eşleştirilir. Aday eşleşmeler frekans tutarlılığı ve zamansal ayrıklık kısıtlarıyla süzülür; tanımlayıcı benzerliği ve yama düzeyi NCC doğrulamasıyla yalancı eşleşmeler elenir. Doğrulanan eşleşmelerin zaman ofsetinde kümelenmesi histogram oylama ile baskın kaymayı verir ve bu kayma üzerinden sahte segment sınırları lokalize edilir. Yerel doku-temelli eşleştirme ve çok aşamalı doğrulama yapısı sayesinde yöntem, bozulma koşullarında dahi tekrar eden spektral örüntüleri koruyarak ataklara karşı dayanıklı bir tespit/lokalizasyon profili sunar. Yöntem, TIMIT-ACMF, Arabic-ACMF ve Türkçe-ACMF veri setlerinde ataksız ve farklı saldırı koşullarında doğruluk ve F-ölçümü ile değerlendirilmiş; en yüksek sonuçlar Arabic-ACMF (ataksız) %92,50 / %92,50, TIMIT-ACMF (ataksız) %90,62 / %89,86 ve Türkçe-ACMF (medyan filtreleme) %90,64 / %91,41 olarak elde edilmiştir.

Keywords

ACMF, Ses adli analizi, Spektrogram, Derin anahtar nokta, XFeat

Audio Copy-Move Forgery Detection with High-Resolution Spectrograms and XFeat Deep Features

Abstract

Audio copy-move forgery is performed by pasting a segment taken from the same audio recording back into the same recording at a different time interval. Because the forgery is performed within the same recording, the source/target segment having the same channel and background characteristics makes forgery detection quite difficult. For this purpose, in the proposed study, the signal is converted into high-resolution spectrogram images, and repetitive time-frequency patterns are matched with XFeat-based deep local descriptors. Candidate matches are filtered with frequency coherence and temporal discreteness constraints; pseudo-matches are eliminated with descriptor similarity and patch-level NCC verification. The clustering of verified matches in time offset gives the dominant shift with histogram voting, and the boundaries of the forged segment are localized based on this shift. Thanks to the local texture-based matching and multi-stage verification structure, the method offers a robust detection/localization profile against attacks by preserving repetitive spectral patterns even under degradation conditions. The method was evaluated using accuracy and F-measure in TIMIT-ACMF, Arabic-ACMF, and Turkish-ACMF datasets under attack-free and different attack conditions; the highest results were obtained as follows: Arabic-ACMF (attack-free) 92.50% / 92.50%, TIMIT-ACMF (attack-free) 90.62% / 89.86%, and Turkish-ACMF (median filtering) 90.64% / 91.41%.

Keywords

ACMF, Forensic audio analysis, Spectrogram, Key point, XFeat

References

Xiao, J. N., Jia, Y. Z., Fu, E. D., Huang, Z., Li, Y., & Shi, S. P. (2014). Audio authenticity: Duplicated audio segment detection in waveform audio file. Journal of Shanghai Jiaotong University (Science), 19(4), 392-397.
Su, Z., Li, M., Zhang, G., Wu, Q., & Wang, Y. (2023). Robust audio copy-move forgery detection on short forged slices using sliding window. Journal of Information Security and Applications, 75, 103507.
Su, Z., Li, M., Zhang, G., Wu, Q., Li, M., Zhang, W., & Yao, X. (2022). Robust audio copy-move forgery detection using constant Q spectral sketches and GA-SVM. IEEE Transactions on Dependable and Secure Computing, 20(5), 4016-4031.
Yan, Q., Yang, R., & Huang, J. (2015, April). Copy-move detection of audio recording with pitch similarity. In 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 1782-1786). IEEE.
Yan, Q., Yang, R., & Huang, J. (2019). Robust copy–move detection of speech recording using similarities of pitch and formant. IEEE Transactions on Information Forensics and Security, 14(9), 2331-2341.
Imran, M., Ali, Z., Bakhsh, S. T., & Akram, S. (2017). Blind detection of copy-move forgery in digital audio forensics. IEEE Access, 5, 12843-12855.
Mannepalli, K., Krishna, P., Krishna, K., & Krishna, K. R. (2019). Copy and move detection in audio recordings using dynamic time warping algorithm. International Journal of Innovative Technology and Exploring Engineering, 9(2), 2244-2249.
Ustubioglu, B., Küçükuğurlu, B., & Ulutas, G. (2022). Robust copy-move detection in digital audio forensics based on pitch and modified discrete cosine transform. Multimedia Tools and Applications, 81(19), 27149-27185.
Huang, X., Liu, Z., Lu, W., Liu, H., & Xiang, S. (2019). Fast and effective copy-move detection of digital audio based on auto segment. International Journal of Digital Crime and Forensics (IJDCF), 11(2), 47-62.
Peng, K., Shi, C., Li, X., Qi, C., Qiu, X., & Niu, X. (2025). Robust audio copy-move detection and localisation based on Smoothed Pseudo Wigner-Ville Distribution and adaptive processing. Applied Acoustics, 231, 110429.

Akdeniz, F., & Becerikli, Y. (2024). Detecting audio copy-move forgery with an artificial neural network. Signal, Image and Video Processing, 18(3), 2117-2133.
Ustubioglu, A., Ustubioglu, B., & Ulutas, G. (2023). Mel spectrogram-based audio forgery detection using CNN. Signal, Image and Video Processing, 17(5), 2211-2219.
Dincer, S., Ustubioglu, B., Ulutas, G., Tahaoglu, G., & Ustubioglu, A. (2023, September). Robust audio forgery detection method based on capsule network. In 2023 International Conference on Electrical and Information Technology (IEIT) (pp. 243-247). IEEE.
Ustubioglu, B. (2024). An attack-independent audio forgery detection technique based on cochleagram images of segments with dynamic threshold. IEEE Access, 12, 82660-82675.
Ulutas, G., Ustubioglu, A., Ustubioglu, B., & Tahaoglu, G. (2022, July). Localization of Forgery on Audio Clips Using GLCM Features and Mel Spectograms. In 2022 45th International Conference on Telecommunications and Signal Processing (TSP) (pp. 314-317). IEEE.
Ustubioglu, B., Tahaoglu, G., & Ulutas, G. (2023). Detection of audio copy-move-forgery with novel feature matching on Mel spectrogram. Expert Systems with Applications, 213, 118963.
Ulutas, G., Tahaoglu, G., & Ustubioglu, B. (2022, July). Forge audio detection using keypoint features on mel spectrograms. In 2022 45th International Conference on Telecommunications and Signal Processing (TSP) (pp. 413-416). IEEE.
Ustubioglu, B., Tahaoglu, G., Ulutas, G., Ustubioglu, A., & Kilic, M. (2024). Audio forgery detection and localization with super-resolution spectrogram and keypoint-based clustering approach. The Journal of Supercomputing, 80(1), 486-518.
Ustubioglu, B., Tahaoglu, G., Ustubioglu, A., Ulutas, G., & Kilic, M. (2025). A novel audio copy move forgery detection method with classification of graph-based representations. IEEE Access.
Garofolo, J. S., Lamel, L. F., Fisher, W. M., Pallett, D. S., Dahlgren, N. L., Zue, V., & Fiscus, J. G. (1993). TIMIT acoustic-phonetic continuous speech corpus.
Potje, G., Cadar, F., Araujo, A., Martins, R., & Nascimento, E. R. (2024). Xfeat: Accelerated features for lightweight image matching. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 2682-2691).
Ustubioglu, B., Tahaoglu, G., Ayaz, G. O., Ustubioglu, A., Ulutas, G., Cosar, M., ... & Kılıc, M. (2024, July). KTUCengAudioForgerySet: a new audio copy-move forgery dataset. In 2024 47th International Conference on Telecommunications and Signal Processing (TSP) (pp. 123-129). IEEE.
Halabi, N. (2016). Arabic speech corpus. Oxford text archive core collection.
Ozgen, E., & Altay, S. Y. (2025). Detecting Audio Copy-Move Forgeries on Mel Spectrograms via Hybrid Keypoint Features. Applied Sciences, 15(21), 11845.
Buades, A., Coll, B., & Morel, J. M. (2005, June). A non-local algorithm for image denoising. In 2005 IEEE computer society conference on computer vision and pattern recognition (CVPR'05) (Vol. 2, pp. 60-65). Ieee.
Nam, J., Mysore, G. J., Ganseman, J., Lee, K., & Abel, J. S. (2010, September). A super-resolution spectrogram using coupled PLCA. In INTERSPEECH (pp. 1696-1699).
Wang, F., Li, C., & Tian, L. (2017, October). An algorithm of detecting audio copy-move forgery based on DCT and SVD. In 2017 IEEE 17th International Conference on Communication Technology (ICCT) (pp. 1652-1657). IEEE.

Details

Primary Language

Turkish

Subjects

Computer Software

Journal Section

Research Article

Authors

Arda Üstübioğlu ^*
0000-0002-8656-8697
Türkiye

Publication Date

March 8, 2026

Submission Date

December 30, 2025

Acceptance Date

February 2, 2026

Published in Issue

Year 2026 Volume: 16 Number: 1

DOI

https://doi.org/10.31466/kfbd.1852460

IZ

https://izlik.org/JA85WD32BM

APA

Üstübioğlu, A. (2026). Yüksek Çözünürlüklü Spektrogramlar ve XFeat Derin Öznitelikleri ile Ses Kopyala-Taşı Sahteciliği Tespiti. Karadeniz Fen Bilimleri Dergisi, 16(1), 432-453. https://doi.org/10.31466/kfbd.1852460

AMA

1.Üstübioğlu A. Yüksek Çözünürlüklü Spektrogramlar ve XFeat Derin Öznitelikleri ile Ses Kopyala-Taşı Sahteciliği Tespiti. KFBD. 2026;16(1):432-453. doi:10.31466/kfbd.1852460

Chicago

Üstübioğlu, Arda. 2026. “Yüksek Çözünürlüklü Spektrogramlar Ve XFeat Derin Öznitelikleri Ile Ses Kopyala-Taşı Sahteciliği Tespiti”. Karadeniz Fen Bilimleri Dergisi 16 (1): 432-53. https://doi.org/10.31466/kfbd.1852460.

EndNote

Üstübioğlu A (March 1, 2026) Yüksek Çözünürlüklü Spektrogramlar ve XFeat Derin Öznitelikleri ile Ses Kopyala-Taşı Sahteciliği Tespiti. Karadeniz Fen Bilimleri Dergisi 16 1 432–453.

IEEE

[1]A. Üstübioğlu, “Yüksek Çözünürlüklü Spektrogramlar ve XFeat Derin Öznitelikleri ile Ses Kopyala-Taşı Sahteciliği Tespiti”, KFBD, vol. 16, no. 1, pp. 432–453, Mar. 2026, doi: 10.31466/kfbd.1852460.

ISNAD

Üstübioğlu, Arda. “Yüksek Çözünürlüklü Spektrogramlar Ve XFeat Derin Öznitelikleri Ile Ses Kopyala-Taşı Sahteciliği Tespiti”. Karadeniz Fen Bilimleri Dergisi 16/1 (March 1, 2026): 432-453. https://doi.org/10.31466/kfbd.1852460.

JAMA

1.Üstübioğlu A. Yüksek Çözünürlüklü Spektrogramlar ve XFeat Derin Öznitelikleri ile Ses Kopyala-Taşı Sahteciliği Tespiti. KFBD. 2026;16:432–453.

MLA

Üstübioğlu, Arda. “Yüksek Çözünürlüklü Spektrogramlar Ve XFeat Derin Öznitelikleri Ile Ses Kopyala-Taşı Sahteciliği Tespiti”. Karadeniz Fen Bilimleri Dergisi, vol. 16, no. 1, Mar. 2026, pp. 432-53, doi:10.31466/kfbd.1852460.

Vancouver

1.Arda Üstübioğlu. Yüksek Çözünürlüklü Spektrogramlar ve XFeat Derin Öznitelikleri ile Ses Kopyala-Taşı Sahteciliği Tespiti. KFBD. 2026 Mar. 1;16(1):432-53. doi:10.31466/kfbd.1852460

The published articles in The Black Sea Journal of Sciences are licensed under Creative Commons Attribution-NonCommercial 4.0 International

Yüksek Çözünürlüklü Spektrogramlar ve XFeat Derin Öznitelikleri ile Ses Kopyala-Taşı Sahteciliği Tespiti

Abstract

Keywords

Audio Copy-Move Forgery Detection with High-Resolution Spectrograms and XFeat Deep Features

Abstract

Keywords

References

Details

Primary Language

Subjects

Journal Section

Authors

Publication Date

Submission Date

Acceptance Date

Published in Issue

DOI

IZ

Cite