Research Areas
The audio and acoustics area covers a broad range of topics around studio engineering, electroacoustics, room acoustics, digital audio processing, psychoacoustics, and music perception. Research in the lab’s audio and acoustics area has traditionally a tight connection to the study programme ‘Electrical Engineering and Audio Engineering’ that is run jointly by Graz University of Technology and Graz University of Music and Performing Arts. Current work in the lab addresses questions such as: How can we enhance music perception of listeners with hearing loss? What are critical acoustical features for the parsing of sound mixtures and the perception of musical instrument sounds? How to develop good audio-tactile interfaces for enhanced audio perception? Through the combination of methods from audio processing, psychoacoustics, and data modeling, we seek to understand of how listeners make sense of our blooming and buzzing acoustic reality.
An intelligent system is able to perceive, learn, reason, and act in a prudent way. This involves various perception modalities such as input from cameras, microphones, sonar and other more exotic sensors. Furthermore, machine learning and pattern recognition techniques are important ingredients for reasoning under uncertainty in intelligent systems. One major aim is to extract relevant information from massive data in a semi-automatic fashion using computational and statistical methods. This interdisciplinary research is related to many fields throughout science and engineering, i.e., statistics, probability, and graph theory, optimization methods, logic, speech and image processing, control theory etcetera. The focus is on providing solutions for tasks where some kind of intelligence is inevitably essential. Application areas include bioinformatics, computer vision, natural language processing, speech processing, man-machine interfaces, expert systems, and robotics amongst others.
Nonlinear Signal Processing is an emerging discipline combining knowledge from signal processing, adaptive systems, nonlinear dynamical systems, statistics and information theory, computation, and mixed-signal processing systems realization. Nonlinearity shows itself as a curse in many physical system realizations where analog effects may deviate strongly from their idealized linear behavior. The modeling of these nuisance effects and their adaptive digital compensation is a key application of nonlinear signal processing in the realm of mixed-signal processing systems such as power amplifiers for wire-line and wireless communications. Nonlinearity can also be a blessing when it comes to the modeling, compression, and interpretation of information sources where deterministic nonlinear dynamics rivals with conventional statistical models in representing randomness of a source, i.e., its innovation or information content. This gives rise to more accurate signal models and to an understanding of information flow in computational algorithms and distributed signal processing networks. This research area strongly interacts with the art of algorithm engineering as an extension to circuit and system design, including the mapping on reconfigurable architectures.
Speech Communication covers speech production and speech perception of the sounds used in spoken human language. It is a highly interdisciplinary field that is studied by several academic disciplines including acoustics, psychology, speech pathology, linguistics, cognitive science, communication studies, computer science, and signal processing. Research in Speech Communication at our lab focuses on automatic speech processing techniques for human-machine-interaction, for enhancing speech transmission, and for improving life quality of disabled persons. This involves research topics in speech signal processing, like speech analysis, speech enhancement and transmission, as well as research topics in automatic speech communication, like acoustic source localization, speech and speaker recognition, and several language technologies. To this end, our lab provides the Speech Communication research area with an ideal combination of other research areas, like nonlinear signal processing, and intelligent systems.
Wireless Communications is a multidisciplinary subject, requiring expertise in radio propagation, antennas, RF electronics, analog and digital (mixed-signal) signal processing, systems theory, some control theory, and many mathematical topics, as statistics, coding theory, queuing theory, game theory, and others. Continuously, new topics have been emerging over the past years. Only a few years back, concepts as orthogonal frequency division multiplexing (OFDM), multi-input multi-output (MIMO) systems, ultrawideband (UWB) systems, cooperative communications, or cognitive radio have been considered as visionary but most likely unpractical ideas. Nowadays, many of those have found applications. They have been included in communications standards and they are being discussed in lecture courses on wireless communications. Ultra-wideband (UWB) systems and techniques have become our key topic, potentially offering ultra-low-power wireless links that are robust against the ever-present multipath propagation in wireless channels. Based on UWB techniques, high-accuracy indoor localization, next-generation RFID systems, and ultra-wideband channel modeling have been the main research topics in Wireless Communications during the last years.