Sidharth

I am a CSE PhD student in the Soundability Lab at the University of Michigan, advised by Dhruv Jain and working with Hao-Wen (Herman) Dong. I work on speech and audio AI that holds up in the real world: noisy rooms, overlapping talkers, and devices people actually wear.

My current focus is enrollment-free target speech extraction: picking out the person you are talking to without first asking them for a clean voice sample. In Unmixing the Crowd, I learn persistent speaker representations directly from mixtures. Before Michigan, I built real-time speech enhancement at BrainChip and Skyworks, and this summer worked on spatial audio and sound localization at Bose Research.