I completed my PhD in Electrical & Computer Engineering at Johns Hopkins University in the Center for Language and Speech Processing, co-advised by Sanjeev Khudanpur and Shinji Watanabe, having attended Carnegie Mellon University for my undergraduate degree prior to that. After spending a few years as a Research Scientist on Amazon’s acoustic event detection team, I returned to Hopkins as a research scientist. My areas of expertise are in deep learning, statistical modeling, and signal processing. In general, I like to work on waveform-level processing of acoustic signals.
In particular, I am interested in working with conversations captured in ambient acoustic environments, featuring long-term recordings picking up multiple far-field sources. Processing the speech in such recordings can be challenging due to comparatively low volume, often degraded with interfering sounds, reverberation, or simply other speech. One of my primary focuses is in the enhancement of the speech signals. My PhD dissertation focused on single-channel speech separation, the task of producing a single audio waveform for each speaker in a recording where mulitple people are speaking simultaneously. I am also interested in speaker diarization and speaker identification.