In this thesis we studied large-vocabulary continuous speech recognition. We considered the components necessary to realise a large-vocabulary speech recogniser and how systems such as Sphinx and HTK solved the ...

Statistical speech recognition systems typically utilise a set of statistical models of subword units based on the set of phonemes in a target language. However, in continuous speech it is important to consider co-articulation ...

ENGLISH ABSTRACT: Phoneme recognition systems typically suffer from low classification accuracy. Recognition for South African English is especially difficult, due to the variety of vastly different accent groups. This ...