Department of Physics

Physics Ph.D. Defense - Shrishti Yadav.

 

 

 

Signals don’t always speak loudly—sometimes they hide beneath layers of competing sound. In this talk, I present a new approach to detecting subtle acoustic signals that are masked by other sources in complex, mixed sound fields. Instead of filtering or relying on known templates, I turn to tools from chaos theory—Recurrence Plots and Poincaré Maps—to transform raw audio
into visual landscapes where hidden structure becomes visible. What’s novel is the use of multi-sensor microphone data to drive these phase space representations. By conditioning the data across multiple channels, subtle but consistent features in the signal begin to emerge—features that single-sensor methods or standard signal processing may miss. I’ll walk through the signal processing pipeline, explain how these plots are constructed, and show results from real acoustic experiments where this approach reveals signals deeply embedded in noise.