Author(s)

Ping Xia

Affiliation(s)

School of Engineering, Guangzhou College of Technology and Business, Foshan 528138, Guangdong, China.

Corresponding Author

Ping Xia

ABSTRACT

This paper proposes a theoretical framework for DenseNet-based voiceprint recognition, which incorporates spectrogram enhancement and adaptive histogram equalization to overcome the limitations of conventional methods in feature extraction robustness under noisy conditions. The framework synergistically combines spectral feature enhancement with DenseNet's dense connectivity, achieving both improved feature discriminability and deep feature reuse through: optimized time-frequency representation via enhanced spectrograms, hierarchical feature propagation enabled by dense blocks. Theoretical analysis confirms the framework's capability to maintain recognition stability against acoustic interference, establishing a novel biometric authentication paradigm for cybersecurity situational awareness systems.

KEYWORDS

Voiceprint recognition; Spectrogram feature enhancement; Histogram equalization; Cybersecurity; Situational awareness

CITE THIS PAPER

Ping Xia. A theoretical architecture of voiceprint recognition for network security situational awareness. World Journal of Information Technology. 2025, 3(2): 31-36. DOI: https://doi.org/10.61784/wjit3031.

REFERENCES

[1] Alam M J, Kenny P, Ouellet P, et al. Multi-task learning for speaker verification and antispoofing using Gaussian mixture models. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2020, 28: 1696-1709.

[2] Villalba J, Chen N, Snyder D, et al. State-of-the-art speaker recognition with neural network embeddings in NIST SRE18 and Speakers in the Wild evaluations. Computer Speech & Language, 2021, 60: 101026.

[3] Ferrer L, McLaren M, Lawson A. Probabilistic linear discriminant analysis with vector embeddings for speaker verification. IEEE Journal of Selected Topics in Signal Processing, 2021, 15(4): 1029-1042.

[4] Snyder D, Garcia-Romero D, Sell G, et al. X-vectors: Robust DNN embeddings for speaker recognition. IEEE Transactions on Audio, Speech, and Language Processing, 2018, 26(7): 110-119.

[5] Chung J S, Nagrani A, Zisserman A. VoxCeleb2: Deep speaker recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 13(5): 532-541.

[6] Villalba J. Advanced speaker recognition using deep neural networks. Carnegie Mellon University, 2020.

[7] Hajibabaei M, Dai D. Unified hypersphere embedding for speaker recognition. IEEE Transactions on Audio, Speech, and Language Processing, 2019, 45(3): 321-330.

[8] Liu X, Ji Y, Liu C. Voiceprint recognition based on LSTM neural network . Computer Science, 2021，48(S2), 270-274.

[9] Wang H P. Speaker recognition based on deep bidirectional LSTM network. Computer Engineering and Design, 2020, 41(06): 1768-1772.

[10] Zhao H, Yue L, Wang W, et al. Research on end-to-end voiceprint recognition model based on convolutional neural network. Journal of Web Engineering, 2021, 20(5): 1573-1586.

[11] Yan H, Dong Y, Wang P, et al. Research on voiceprint recognition based on CNN-LSTM network. Computer Application and Software, 2019, 36(04): 166-170.

[12] Guo D, Zhou Q. Land-air call voiceprint recognition based on residual neural network. Modern Computer, 2020(07): 9-13.

[13] Liu Y, Liang H, Liu G, et al. Voiceprint recognition method based on ResNet-LSTM. Computer System Applications, 2021, 30(06): 215-219.