No label defined (Q5287542)

[The articles of this volume will not be indexed individually.] The book surveys major topics of the methodology used for speech recognition, speech understanding and speech synthesis, including several applications in speech processing systems. Most of the articles appeared in 1989 in an abridged and earlier version in two issues of the journal ``Informationstechnik -- it''. The contributions on methodological aspects of speech processing treat major issues of acoustical and linguistic processing. The first of these two topics is covered by articles dealing with stochastic models and algorithms underlying isolated word and continuous speech recognition as well as speaker-dependent and speaker-independent word recognition, the generation of articulation variants, a model of syllable-oriented speech recognition based on a psychoacoustic model of the ear, and a neural network approach to isolated word recognition. The articles on linguistic processing treat approaches to bi-directional chart parsing (island parsing) of spoken language, syntactic and semantic analysis of spoken natural language utterances based on a dialogue model, anaphora resolution, and pragmatic issues of information seeking dialogues. Several speech understanding systems for the German language are introduced: the SPICOS II system for phoneme-based, large-vocabulary continuous speech recognition (applying a stochastic language model for word recognition, grammar-based linguistic models for parsing and semantic interpretation, and a specialized component for dialogue control), the EVAR system with emphasis on its advanced semantic and dialogue processing capabilities, and the RELEXIS text-to-speech synthesis system which automatically generates spoken language output from linear, written text input. Finally, various articles treat the application dimension of speech technology in the framework of man- machine communication based on spoken language, in particular, for manufacturing, office and telecommunication tasks and natural language access to informtion systems.