Syllable structure parsing for continuous speech recognition

A probabilistic syllable parsing algorithm used to extract an underlying syllable structure from acoustic speech realizations is described. In the algorithm, phonological objects are represented in terms of features, syllable positions, and distinctive features, and they are organized within hierarchical structures using a probabilistic function. The algorithm parses the syllable structure from the acoustic speech realizations by applying the probabilistic acoustic-phonological constraints and the probabilistic collocational restrictions involved in the internal constituent features. Performance results for 15 test sentences spoken by 5 male speakers indicate that phonemes are recognized at 90.5% accuracy, and that the syllable structure is parsed at 79.7% accuracy