Topic Four

TO4. In the fourth thematic area we will focus on the analysis and design of new approaches in the area of recognition and synthesis of speech signals. We will concentrate on the robustness in real environments with interference, the analysis and design methods for extracting of paralinguistic information from the speech and analysis and design of the approaches for cooperative communication in the multi-agent environment (people, computers and robots). The main goals of the research will be the following:

TO4.1 Analysis and design methods leading to the robust capture of the speech signal over a greater distance (long distance recognition) by using multi-channel methods and microphone arrays.

Investigator: LRMTT

Output:

TO4.2 Analysis and design methods for speech signal preprocessing and training of acoustic models in order to increase the robustness in noisy environments. The main subjects of the analysis will be methods for elimination of noise and increasing the recognition accuracy in noisy environments as well as the analysis and selection of an optimal type of the parameters of acoustic vectors.

Investigator: LRMTT

Output:

TO4.3 Analysis and design methods of extraction of the paralinguistic information from the speech signal such as information about the speaker's emotions or type of the statement (question/command), etc., creating the potential for adapting the human-machine interactions, because paralinguistic level provides a large amount of useful information that have certain communication tasks.

Investigator: LRMTT

Output:

TO4.4 Analysis and design methodology for cooperative management in multi-agent environment like people, computers and robots, where the communication adapts to the human and assists in solving of the tasks requiring collaboration with computer systems and robots. Analysis of the different strategies and models of dialogue with more than two participants.

Investigator: LRMTT

Output:

TO4.5. Hardware, software, emotional, communicational and implementation aspects of the robotic objects speech output.

Investigator: LRMTT

Output: