PhD Defense: Human-centric Acoustics: The Interplay of Opportunities and Security

Acoustic signals play a fundamental role in human life, serving as a unique medium for communication, interaction, and perception. Beyond speech, our bodies generate various sounds that carry valuable information, from health indicators to environmental cues. These signals not only shape our daily interactions but also offer unprecedented opportunities to enhance user experiences in diverse contexts, whether by improving communication, monitoring health, or safeguarding privacy. This dissertation delves into the intricate interplay of these opportunities and the security challenges inherent in human-centric acoustics, advancing the field toward innovative and user-focused solutions.The contributions of this dissertation span five key areas: enhancing speech communication, enabling continuous ear health monitoring, ensuring the integrity of live speech, protecting user privacy in voice communication, and enabling secure acoustic interaction. To improve communication, this work introduces a spatial filtering algorithm that leverages the harmonic structure of human speech to isolate the speaker’s voice, significantly enhancing call quality in noisy environments using commodity earbuds. For health monitoring, it presents a method to trigger distortion product otoacoustic emissions (DPOAEs) using single-speaker commercial earbuds, enabling accessible, medical-grade ear assessments without specialized equipment. To ensure speech integrity, this research proposes a robust verification system that embeds cryptographic meta-information into live speech, allowing detection of tampering in open recording environments. To safeguard privacy in everyday conversations, it presents a microphone module that prevents automated speaker identification and speech recognition while preserving naturalness for human listeners. Finally, this dissertation introduces an acoustic handwriting tracking system that achieves sub-millimeter precision, enabling secure biometric authentication through sound-based input on voice-enabled devices.These advancements are built on a strong foundation of theoretical principles in signal processing, spatial sensing, human speech production and perception, and machine learning. By seamlessly integrating these disciplines, this work demonstrates the potential for acoustic technologies to address real-world challenges. By addressing the dual challenges of opportunity and security, this research bridges the gap between technological innovation and user-centric needs. It lays a foundation for future acoustic systems that enhance the quality of life while maintaining trust, privacy, and security in an increasingly interconnected world.