Research objective

Speech and language represent the most natural means of human communication. Human-computer interaction was traditionally driven by the computer technology and not by the humans' natural abilities of interaction. Speech and language processing aim at both increasing the naturanless  and efficiency of human-computer interaction, and at enhancing computer-mediated human-human communication. These represent the broad objectives of our research within the Speech and Language Processing Laboratory . In approaching these goals, we are interested in both science and technology of speech and language. 

From the speech science perspective, we are interested in both speech perception and speech production. We believe that more research in speech science is necessary to help new developments in speech technology which currently, in many areas, does not provide results comparable to those of human performance. Our main speech research areas of interest are: speech analysis, speech synthesis, automatic speech recognition, speech coding and speech enhancement. 

Although speech processing represents the main focus of our research, making the human-computer interaction more natural, efficient and intelligent requires research at higher levels of human-communication, such as language. Language processing, thus represents a necessary component that complements speech processing in order to build natural and reliable human-computer dialogue systems. A computer dialogue system requires a variety of speech and language technologies, such as, automatic speech recognition, natural language processing, language understanding,  dialogue management, language generation, and speech synthesis. The new generations of dialogue systems need also to have learning capabilities in order to adapt their language knowledge at both syntactic and semantic levels from interaction with users. They also need to be adaptive to users' preferences, capable to be personalized by their users and to recognize users by their voices, and capable to remember previous interactions with users. Some of these challanges represent our main research objectives in language processing.