Teaching
Current Classes
- Sound Recognition (Ma)
- The focus of this course is on Auditory Cognition and Real-World Sound Recognition. This focus is extremely broad: it ranges from low level physics, associated with sound production and transmission to the questions associated with (ambient) awareness and the intelligent (sound based) interaction with the world, and with very subjective things like annoyance and pleasure. One of the most interesting and essential aspects of audition is that the whole process from sound generation to a purposeful reaction to the content of the sounds occurs with a delay in the order of a second or less, irrespective the sound and the context. This entails that the purposeful integration of essential features of cognition is just as essential as the function of the different subprocesses that are required for the job. This course is a combination of theory and practical work: you have to build a sound source recognizer for uncontrolled envrionments.
- Perception (Ma)
- This course is aimed to provide an overview of the activities of scientific practice. This varies from scientific practices like such as performing experiments, writing papers, peer reviewing to some aspects of scientific policy development. The course uses the topic of Perception as means for this main purpose. In particular it aims to challenge naive ideas about perception and to provide an overview of modern scientific views of the role of perception for cognition and autonomous systems research. In addition it aims to train (critical) reading skills necessary for mastering scientific literature. The theme of the 2010-2011 was: Perception and action are for core affect regulation.
- Taal en Spraaktechnology (Ba, in Dutch)
- Eerst wordt een flink deel van de (spraak-) signaalanalyse in een rap tempo behandeld. Nadruk hierbij ligt op praktische aspecten: waar en hoe informatie te vinden over bepaalde frequenties, hoe te rekenen met tijden en frequenties en op welke wijze goede geluidsopnamen te maken en kwantitatief te analyseren. Centraal hierbij staat de Fast Fourier Transform (FFT). Daarna wordt de standaard spraakherkenningstechnologie besproken aan de hand van een modern commercieel dicteersysteem (Dragon NaturallySpeaking) op basis van HMM-technologie. Dit systeem wordt getraind en in verschillende situaties getest. Hierbij wordt een aantal overeenkomsten en verschillen tussen het menselijk spraakverwerkingssysteem en de technische uitvoering ervan onderzocht. Het vak wordt afgesloten door in te gaan op een aantal cognitieve aspecten van spraak en spraaksignaalanalyse.