Learning prosodic sequences using the fundamental frequency variation spectrum
2008 (English)In: Proceedings of the Speech Prosody 2008 Conference, Campinas, Brazil: Editora RG/CNPq , 2008, 151-154 p.Conference paper (Refereed)
We investigate a recently introduced vector-valued representation of fundamental frequency variation, whose properties appear to be well-suited for statistical sequence modeling. We show what the representation looks like, and apply hidden Markov models to learn prosodic sequences characteristic of higher-level turn-taking phenomena. Our analysis shows that the models learn exactly those characteristics which have been reported for the phenomena in the literature. Further refinements to the representation lead to 12-17% relative improvement in speaker change prediction for conversational spoken dialogue systems.
Place, publisher, year, edition, pages
Campinas, Brazil: Editora RG/CNPq , 2008. 151-154 p.
Computer Science Language Technology (Computational Linguistics)
IdentifiersURN: urn:nbn:se:kth:diva-51959ScopusID: 2-s2.0-84902655943OAI: oai:DiVA.org:kth-51959DiVA: diva2:465249
[SP-2008] Speech Prosody 2008, Fourth International Conference, Campinas, Brazil, May 6-9, 2008
tmh_import_11_12_14 QC 201112212011-12-142011-12-142011-12-21Bibliographically approved