David van Kuijk

Learn More
The acoustic realization of vowels with lexical stress generally differs substantially from their unstressed counterparts, which are more reduced in spectral quality, shorter in duration, weaker in intensity and tend to have a flatter spectral tilt. Therefore, in an automatic speech recognizer it would appear profitable to train separate models for the(More)
The acoustic realization of vowels with lexical stress generally differs substantially from their unstressed counterparts , which are more reduced in spectral quality, shorter in duration, weaker in intensity and tend to have a flatter spectral tilt. Therefore, in a continuous speech recognizer (CSR) it would appear profitable to train separate models for(More)
For both human and automatic speech recognizers it is difficult to segment continuous speech into discrete units such as words. Word segmentation is so hard because there seem to be no self-evident cues for word boundaries in the speech stream. However, it has been suggested that English listeners can profit from the occurrence of full vowels (i.e. vowels(More)
A new psycholinguistically motivated and neural network based model of human word recognition is presented. In contrast to earlier models it uses real speech as input. At the word layer acoustical and temporal information is stored by sequences of connected sensory neurons which pass on sensor potentials to a word neuron. In experiments with a small lexicon(More)
In recent years computational models have become more and more important in testing processing mechanisms assumed to underlie human spoken-word recognition. Models like TRACE (McClelland & Elman, 1986) and Shortlist (Norris, 1994) have given us much insight in the effects of, for instance, competition between words in the mental lexicon and the use of(More)
  • 1