Specific features of modern voice protection systems

• Published in: Bezopasnostʹ informat͡s︡ionnykh tekhnologiĭ Vol. 24; no. 4; pp. 71 - 79
• Main Author: Ustinov, Roman A.
• Format: Journal Article
• Language: English
• Published: Joint Stock Company "Experimental Scientific and Production Association SPELS 01.11.2017
• Subjects: adaptive compression algorithms; audiomarker; protection of speech information; speech intelligibility; threats to information security
• ISSN: 2074-7128; 2074-7136
• DOI: 10.26583/bit.2017.4.08

Nowadays, speech technologies are among the most vibrant sectors of the world’s economy. Of high importance is the problem of ensuring the security of speech information (SI). Here we discuss SI protection systems within a modern communication model. The model is multimodal, multithreaded, and implies a large number of subscribers interacting via several communication lines. With this in mind, we perform a detailed analysis of threats to the confidentiality, integrity and accessibility of SI. Existing methods of counteraction against these threats are discussed, and shown to be insufficient to ensure the safety of voice messages (VM) in full. Mean while, there are new threats to the integrity and accessibility of SI, the solutions for which are either do not exist, or only being developed. We propose our original approach to counter these threats. Steganography methods are the most promising for ensuring the integrity of the VM.  In particular, using audiomarkers allows one to reliably trace speaker’sidentity throughout the entire communication session. In order to counter the threats to SI availability due to the capacity of the communication channel and the limited volumes of VM data storage, it is necessary to improve existing adaptive speech compression algorithms, along with developing new ones. Furthermore, such algorithms must keep the specified level of speech intelligibility.