The full text of this item is not available at this time because the student has placed this item under an embargo for a period of time. The Libraries are not authorized to provide a copy of this work during the embargo period, even for Texas A&M users with NetID.
An Exposition of Speech-Based Threats in the Face of Emerging Technology
Abstract
Speech privacy is a lasting concern because of the many scenarios in which sensitive information is spoken out loud. Such scenarios can include discussion of business secrets, medical information, banking information, or even personal details of a person that should remain private. Research has shown that both the audio and vibration domains can be utilized to compromise this information. With the rapid development of new signal processing tools, sensor technology, and mobile devices that house multiple sensor types and often become ubiquitous, the potential threat to speech privacy only continues to grow. Additionally, the popularity of voice-controllable systems has introduced new avenues for attacks that could prove detrimental to their users.
In this dissertation, we reveal new perspectives on speech-based threats as they relate to advanced and emerging technologies. We consider both passive threats to speech privacy and active threats against voice-controllable systems. Our work involves multiple aspects of signal process-ing and analysis to reveal the feasibility of speech-based threats as they apply to modern day attack scenarios. We utilize state-of-the-art sensing equipment and processing techniques to understand when such attacks can be executed and what sensitive information can be compromised. Addition-ally, our work was conducted to study the threats as a prerequisite to inform the design of future defensive solutions and mitigating techniques.
Our dissertation provides the following contributions: First, we perform a security analysis of existing studies that introduce speech privacy attacks in the vibration domain and determine their practicality for application against day-to-day live human speech scenarios. Second, we explore the potential for speech eavesdropping in the vibration domain using both state-of-the-art laser technology available in the commercial space, and new sensing approaches centered around existing technology. Third, we identify a limitation in an emerging speech privacy solution designed for modern smart speaker devices. Fourth, we demonstrate new attacks against smart speaker de-vices in the acoustic domain. And fifth, we use the information gained from this work to discuss approaches to defensive solutions that may withstand the capabilities of these threats.
Citation
Walker, Payton Ross (2022). An Exposition of Speech-Based Threats in the Face of Emerging Technology. Doctoral dissertation, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /198009.