ISCA Archive - The head system and its approach to rule based acoustic-phonetic recognition of speech (original) (raw)

The head system and its approach to rule based acoustic-phonetic recognition of speech

Anders Baekgaard, Paul Dalsgaard, Peter Holtse

This paper presents an approach towards using acoustic-phonetic knowledge in Automatic Speech Recognition (ASR). Signal processing hardware and algorithms are described, and a method for the representation and application of phonetic, phonological, and linguistic knowledge is outlined. Finally, the HEAD system, in which these algorithms are implemented, is presented.