picture taken
by Cecile Kuijpers
photo by
Janienke Sturm
Research associate in the Language and Speech Group at the Radboud University, Nijmegen.
Current research interests
Previous research
Contact
| |
|
|
|
|
| In Pittsburgh, 1996 picture taken by Cecile Kuijpers |
more recent | Nijmegen, 2003 photo by Janienke Sturm |
more recent | current brain wave |
My current research interests are
|
Computational cognition |
In collaboration with Paula Fikkert, Lou Boves, Michele Gubian, Christina Bargmann, Maarten Versteegh, we work on computational models of human learning processes with focus on the early stages of language acquisition. Central focus is to understand human behaviour by developing and testing algorithms that are cognitively plausible in the following sense (a) algorithmic steps relatable to current knowledge about psychological and psycholinguistic research results, e.g. by Smith & Yu (2008) and similar (b) causal stimuli processing in the sense that input data are not observed more than once (c) plausible use of memory resources (b) plausible forms of underlying data processing (incremental, online).
|
|
Medically inspired speech decoding for diagnostic tools |
The use of ASR-based speech decoding techniques to diagnose differences between speech tokens, to support medical research in two directions (a) improvement of adjustments of cochlear implants (FP7 project OPTIFOX, with OtoConsult, Antwerp, Belgium) (b) objective assessment of pathological speech (with Irma Verdonck-de Leeuw and Marieke de Bruijn, VU Medical Center, Amsterdam)
|
| Special topics in speech decoding |
With Heyun Huang: phone recognition based on short 'episodic' segments. With Yang Sun: speeech recognition combining sparse classification (by Jort Gemmeke) and Dynamic Bayesian Networks)
|
|
Speech recognition by humans and by machines |
in collaboration with (among others) Odette Scharenborg. Katrin Kirchhoff and I were guest editors for Speech Communication for a special issue on Bridging the gap between human and automatic speech recognition (Vol. 49, May 2007). |
|
Pronunciation variation
Manifold structure |
New work on the relation between reduction in speech and manifold structure is based on work on pronunication variation and HMM topologies by Annika Hamalainen. In collaboration with Bert Cranen we look into the use of articulatory features and Bayesian Networks for ASR.
|
|
Phonetic detail
Phonetic details in speech |
Work in the context of the European Sound-2-Sense project is carried out by Michele Gubian and Barbara Schuppler.
|
| Mathematics | Click here for a joint paper with Matthijs Coster on the relation between Hamilton walks on the hypercube and Fibonacci matrices. One of the underlying results required has been incorporated into the on-line database of integer sequences (see here). |
| Computational modelling of learning Computational cognition |
ACORNS: Acquisition of Recognition and Communication Skills (FP6, 2006-2009). This project aimed at building a computational model for learning the skills for recognition and communication, based on a small number of elementary characteristics of learning (perception, representation, storage, access/recall, abstraction/generalisation). Partners: Lou Boves, Hugo Van hamme, Roger Moore, Unto Laine, Bastiaan Kleijn and colleagues. |
| Multimodal dialogue | The COMIC project (IST, FP5, 2002-2005) focused on defining generic cognitive models for multimodal interaction (speech and gestures) and to evaluate these in a number of demonstrators. When involved in a multimodal man-machine dialogue, people show very different behaviour with respect to modality switches and cross-modal error correction. See the COMIC-website for publications. |
| ASR | From 1995 until end of 2001, I joined the ASR research group of Lernout & Hauspie Speech Products in Wemmel (Belgium). Among our colleagues were Hugo Van hamme, Filiep Vanpoucke, Filip Van Aalten, Tom Claes, Bart d'Hoore, Kristin Daneels, Philippe Gelin, Martine Lapere, Marc Hogenhout, Frederic Beaugendre, Hong Xu and many others. |
| Vowel systems | In 1991, I defended my PhD thesis on Vowel Systems at the University of Amsterdam (supervisor Louis Pols). The aim was to describe the variation observed in vowel systems across languages in the world by using basic physical principles (minimizing articulatory effort while optimizing perceptual contrast). See e.g. here. More recently, the topic received renewed interest by e.g. Bart de Boer (see pdf). |
Contact: l.tenbosch
let.ru.nl