The Computing Innovation Fellows Project

Posted on: Monday, June 15th, 2009
Broad Research Area: AI / Machine Learning / Robotics / Vision, Other

Research Interests:

My broad research interests are in speech and language processing, with a slant toward combining statistical and machine learning techniques with knowledge about language structure from linguistics and speech science.

My recent work involves several topics:

- Models of the articulators (lips, tongue, etc.) for speech recognition (especially for pronunciation modeling in conversational speech), implemented as graphical models.

- Audio-visual speech recognition, using dynamic Bayesian networks to account for the apparent asynchrony between the audio and visual streams.

- Multi-view learning of feature spaces for various speech tasks, using additional views at training time such as video or articulatory measurements.

- Additional modalities in speech processing besides audio and video. For example, one promising source of information appears to be from ultrasonic “microphones”, in which the motion of the articulators is detected via a Doppler shift in a reflected ultrasonic signal.

In addition to these, I am also interested in exploring similarity-based techniques for speech applications and model-based speech processing (such as denoising or reconstruction of degraded signals).

A note about TTI-Chicago: It is an independent institute dedicated to basic research and graduate education in computer science. We are located on the U. of Chicago campus, but are an independent institute with our own PhD program. The main current areas of research at TTI are machine learning, artificial intelligence and related applications (vision, robotics, speech, natural language), theory of computation, programming languages, computational biology, and scientific computing.

Contact Information:

email obfuscated - click to reveal