Е-Публикации
Технически университет - София

Abstract: Steganography methods hide information in video or audio records. The main disadvantage is the reduced resistance to hidden information disclosure using today effective and fast stenographic analysis. The algorithm proposed in this article is based on generative adversarial neural networks to direct transformation of the input audio information as steganography data. The algorithm is tested as corresponding program application and the examination of created generative steganography audio data confirm the achieved high resistance to unauthorized recovering.

References

1. Ar. Febryan, T. Purboyo, and R. E. Saputraq, "Steganography methods on text, audio, image and aideo: A survey", Computer Science, 2017.
2. R.Tanwar, and M. Bisla,"Audio steganography", International Conference on Reliability, Optimization and Information Technology - ICROIT, India, Feb 6-8 2014.
3. H. Kapadia, H. S. Haribau, and H. Pati, "Audio steganography and security using cryptography", IJCSN International Journal of Computer Science and Network, Volume 4, Issue 2, April 2015.
4. N. Meghanathan, and L. Nayak, "Steganalysis algorithms for detecting the hidden information in image, audio and video cover media", International Journal of Network Security & Its Applications, January 2010.
5. R. Ardila et al., "Common Voice: A Massively-Multilingual Speech Corpus", Proceedings of the twelfth Language Resources and Evaluation Conference, Marseille, France, May, 2020.
6. 1DTransposed convolutional layer. https://www.tensorflow.org/api-docs/python/tf/keras/layers/Conv1DTranspose
7. Ch. Donahue, J. McAuley, and M. Puckette, "Adversarial audio synthesis", Conference paper at ICLR, 2019.
8. Dilated 1D convolutional layer. https://keras.io/api/layers/convolution-layers/convolution1d/

Issue