CASSANDRA was the NWO project on which my PhD was based. CASSANDRA stands for Context-Aware SenSing for AggressioN Detection and Risk Assessment. Goal was to detect human activity, specially aggression, in dynamic environments. Because a single sensormodality cannot always detect this audio and video analysis was performed. The two were combined in order to arrive at a cognitive description of the environment. Currently we are wrapping up by writing a joint journal paper with our partners at the university of Amsterdam.

Related publications

Phoneme recognition

Together with Bea Valkenier we are researching how my work on sound source extraction can be applied to recognize vowels and consonants. We have succesfully recognized vowels and are now broadening our view to consonants.

Related publications

Software implementation

The methods developed during my PhD and some more basic signal processing methods are implemented and optimizes in C. This will result in faster, more efficient code that can run independent of Matlab. The implementation in done by DySi and financed by INCAS3


An important problem in sound source recognition is the lack of extensive, well-annotated datasets. In April of 2009 we started, a to-be-collection of these datasets. Also we are building tools to facilitate the annotation of recordings.

Related publications
See also the project by master student Robert van der Linden

Student projects

An overview of student projects I'm involved in. For new projects see Vacancies on our group site. Links may only be reachable from within AI.

Current master projects

Annotation tool Robert van der Linden
Research database for everyday listening Maarten van Grootel

Current bachelor projects

Audibility thresholds for tones and pulses in noise Eva van Viegen

Other projects

Tools for signalgeneration by Jacolien Tange

Past master projects

Classification of signalcomponents in cochleograms by Pim van Oerle, done November 2006