Doctoral Thesis Proposal Automatic Detection and Classification of Prosodic Events

Speech prosody is a valuable carrier of information. Accents and phrase boundaries have been shown to contribute to syntactic disambiguation, semantic, pragmatic and paralinguistic interpretation, and to convey information about topicality, focus, contrast and information status. This thesis will present and evaluate techniques to detect and classify these prosodic events. The acoustic correlates of accents, phrase boundaries and phrase-final tones will also be examined. Spoken language processing systems have not made widespread use of prosodic information. We hypothesize that access to this information should improve the performance of many SLP applications. To support this, we will present proof-of-concept examples integrating hypothesized prosodic event information into speech synthesis, story segmentation, extractive summarization, and prosody tutoring applications.