About me
Research Staff Member
Speech and Language Algorithms
Algorithms and Engines Group
Research lab: Watson Research Center (Yorktown)
John Hershey was a founding member of the Machine Perception Laboratory at UCSD, where he earned his doctorate in 2004, on the topic of Generataive Models for Perception. There he developed computational models for audio-visual speech perception, speech separation, and three-dimensional face tracking. While in graduate school he did extended internships at Mitsubishi Electric Research Labs and at Microsoft Research in the Machine Learning and Applied Statistics group. In 2004-2005 he was Visiting Researcher at Microsoft Research in the Speech Group, where he worked on directed speech detection, using cameras and microphone arrays to modulate speech recognition and reject background noise. Since 2005 he have been at IBM T.J Watson Research Center, in the Speech and Language Algorithms department. Here he has been working on speech models for separating speech from other speakers and background noise, as well as divergence-based algorithms for clustering, mapping, and converting speech models.
Attached file: john-twinpeaks3.jpg
Last updated 3 Aug 2007