Automatic detection and description of syllabic features in continuous speech