Detection of Steganography in LPC, CELP and MELP Audio Standards Encoder Using LVQ Neural Network

Etezadifar, Pouriya; Talati, Saeed; Hassani Ahangar, Mohammadreza; Mollazade, Mahdi

doi:10.22034/joeds.2024.456906.1043

Detection of Steganography in LPC, CELP and MELP Audio Standards Encoder Using LVQ Neural Network

Document Type : Original Article

Authors

pouriya etezadifar ¹

Saeed Talati ²

Mohammadreza Hassani Ahangar ³

Mahdi Mollazade ⁴

¹ Electrical &amp; Computer Engineering, Imam Hussein Comprehensive University, Tehran, Iran

² PhD Candidate, Faculty of Electrical Engineering Department, Imam Hossein University, Tehran,Iran

³ Professor, Faculty of Electrical Engineering Department, Imam Hossein University, Tehran,Iran

⁴ Assistant Professor, Faculty of Electrical Engineering Department, Imam Hossein University, Tehran,Iran

10.22034/joeds.2024.456906.1043

Abstract

Information security is a critical aspect of Electronic Warfare today and is a major focus for many researchers. Steganography, a secure communication method, involves concealing information within other data or content, with audio data offering a higher capacity for hiding information. This article introduces a novel approach for detecting steganography in three audio encoders: LPC, CELP, and MELP, known for their effectiveness in audio encoders. The Steganography technique discussed here involves embedding data in the least significant bit of the audio media. Audio analysis is carried out using the RUNS test features. Variations are identified and utilized to train a neural network (LVQ) by comparing these features in the cover and Stego audio files. The neural network is then employed for classification, and the proposed algorithm is tested on the audio samples. The proposed classification method demonstrates an average accuracy of 93.59% in detecting steganography, showcasing its effectiveness compared to other existing methods.

Keywords

Steganography

Steganalysis

Detection

Linear Predictive

LVQ Neural Network