Initial Situation

„Ambient Intelligence“ is the main research area of the Technische Universität Kaiserslautern. When talking about Ambient Intelligence we mean ubiquitous systems which are able to conceive and recognize what the user is expecting from the system. In the context of this main research area, the working group “Software Architecture” in cooperation with the working group “Software Engineering” has prepared a prototype based on Jini from Sun, which enables the dynamic integration of mobile heterogeneous devices. Jini provides locating distributed services in a network.
The application, which is realized by the prototype, enables the monitoring of the user-pulse by dint of a sensor and if necessary to present a warning on one of the output devices in the case that the defined upper or lower limit were exceeded. In the actual configuration there are already multiple output services for different devices (LCD-Display, Handy, Laptop, …) existing.
In the context of the Hungary-Cooperation “distributed project group”, the prototype was extended by an output service for voice output at the BUTE (Budapest).

Precise task

In the context of this diploma thesis voice recognition should be integrated as input service in the prototype in cooperation with the BUTE. The input service is used in the considered prototype to define the lower and upper limit of the pulse. Furthermore a general conception of ambient intelligence systems should be a research topic within this thesis. By means of the concrete system there should be considered the question to what extent the quality of voice recognition is depending on stress of the user and whether the quality can be significantly improved by inclusion of information concerning the stress.
In detail there should be processed the following tasks:

• Design and conception of ambient intelligence systems considering as example of a prototype
• Integration of the voice recognition in the prototype
• Analysis and implementation of algorithms to improve the voice recognition under physiological stress
• Conception and prototypical implementation of an intelligent control of the voice recognition which includes the system environment (physiological stress of the speaker and speech variance involved)

Requisites

• Knowledge in the programming languages C and Java
• Knowledge in UML
• Fun in working on literature
• Ideal: Previous knowledge in Jini, knowledge in the area of pattern recognition

Prospects

• Gain insight a practical subject
• Designing a concrete architecture for an application
• Direct research work

Special Features

This diploma thesis should be executed in close cooperation with the BUTE. Thus it is associated with a stay in Budapest / Hungary.