Knowledge-based approaches versus mathematical model based algorithms: the case of speech recognition

A comprehensive overview of the advances in the domain of speech recognition and how self-organizing methods have been used as an alternative to knowledge-based methods is given. Problems related to automatic speech processing are considered. The knowledge-based approaches that have been used to address those problems are described. The author introduces self-organizing approaches, starting from pattern matching techniques, introducing the methodological novelties that allow for progress along three axes: from isolated-word recognition to continuous speech, from speaker-dependent recognition to speaker-independent recognition and from a small vocabulary to a large one. Special emphasis is placed on the improvements made possible by hidden Markov models resulting in progress simultaneously obtained along the different axes, in terms of improved performance for difficult vocabularies, or in more robust systems