Locus Equations Determination Using the SpeechDat(II)

摘要

This paper presents a corpus-based approach to determination of locus equations for Slovenian language. The SpeechDat(II) spoken language database is analyzed first for all available target VCV contexts in order to yield candidate subsets for the acoustic-phonetic measurements. Only the VCVs embedded within judiciously chosen carrier utterances are then selected for the (F2_(vowel), F2_(onset)) measurements. The paper discusses challenges, methodology, and results obtained on the 1000-speaker Slovenian SpeechDat(II) database in the framework of /VbV/, /VdV/, and /VgV/-based determination of locus equations.