I am currently an engineer for the Siri team at Apple in Cupertino, CA.
My background is in speech / audio / music signal processing, pattern recognition, and information retrieval.
I am especially fond of creating mathematical models, implementing them on computers, and applying them to real-world problems.
Some of my past and present work includes:
- Speaker search: I developed a novel method of searching speakers over large populations (10,000 or more) that is orders of magnitude faster than all competing methods. [C1] [P1]
- Speaker diarization: I developed a novel statistical utterance comparison model for speaker diarization. [J1] [C2] [P1]
- Music information retrieval: I developed novel methods of efficiently searching music pitch contours using wavelets and a segmented dynamic time-warping algorithm. [C3] [C4] [P2]
- Auditory modeling for speech processing: For my Ph.D. thesis, I created a theoretical framework for interpreting a computational auditory model and analyzing its noise robustness. [J2] [C6] [C8] [C9].
- Category-dependent feature selection: Also for my Ph.D. thesis, I innovated novel feature selection and pattern recognition methods for automatic speech recognition using high-dimensional auditory data. [C7] [P3].
Contact e-mail: woojay at gmail dot com