University of Toronto and the Toronto Rehabilitation Institute
Frank Rudzicz joined IATSL in the summer of 2011. His postdoctoral research focuses on speech recognition and artificial intelligence in applications designed for individuals with special speech characteristics. For example, Frank is working to build intelligent speech dialogues to help individuals with dementia perform daily tasks as part of the broader COACH project. Other work includes automatic modifications to unintelligible speech signals in order to make them more understandable to typical listeners, and industrial work with Quillsoft Limited in building assistive communication software.
Prior to joining IATSL, Frank was a PhD student in the University of Toronto’s Department of Computer Science. His doctoral thesis, “Production knowledge in the recognition of dysarthric speech” concerned building models for speech recognition that incorporated measurements of the vocal tract for individuals with speech disabilities. For example, Frank significantly improved the rates of correct recognition for speakers with cerebral palsy by writing machine-learning algorithms that learned the statistical relationships between the sounds of speech and the motion of the articulators. The motion of the articulators was measured at relevant points on the lips and tongue using a highly accurate technique called electromagnetic articulography at the Department of Speech-Language Pathology at the University of Toronto. More information on my background can be found on my website.
Rudzicz, F. (in press) Articulatory knowledge in the recognition of dysarthric speech. in IEEE Transactions on Audio, Speech, and Language Processing.
Rudzicz, F., Namasivayam, A.K., Wolff, T. (in press) The TORGO database of acoustic and articulatory speech from speakers with dysarthria. in Language Resources and Evaluation.
Mengistu, K.T., Rudzicz, F. (2011) Adapting acoustic and lexical models to dysarthric speech. Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP11), May 22--27, Prague Czech Republic.
Rudzicz, F. (2010) Learning mixed acoustic/articulatory models for disabled speech. Proceedings of the Workshop on Machine Learning for Assistive Technologies at the twenty-fourth annual conference on Neural Information Processing Systems (NIPS 2010), pages 70--78, December, Whistler, British Columbia.
Reimer, M., Rudzicz, F. (2010) Identifying articulatory goals from kinematic data using principal differential analysis. Proceedings of Interspeech 2010, pages 1608--1611, September 26-30, Makuhari Japan.
Rudzicz, F. (2010) Correcting errors in speech recognition with articulatory dynamics. Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics (ACL 2010), July 11-16, Uppsala Sweden.
Rudzicz, F. (2010) Towards a noisy-channel model of dysarthria in speech recognition. Proceedings of the First Workshop on Speech and Language Processing for Assistive Technologies (SLPAT) at the 11th Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL 2010), June 2-6, Los Angeles California, pages 80--88.
Rudzicz, F. (2010) Adaptive kernel canonical correlation analysis for estimation of task dynamics from acoustics. Proceedings of the 2010 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP'10), March 14-19, Dallas, Texas.