作者: Francis Nolan , Gea de Jong , Toby Hudson , Kirsty McDougall
DOI:
关键词: Phonetics 、 Sound change 、 Speech recognition 、 Structure (mathematical logic) 、 Formant 、 Computer science 、 Dynamics (music) 、 SIGNAL (programming language)
摘要: DyViS is a forensic phonetics research project which investigates dynamic variability in speech from two perspectives: firstly, the use of (timevarying) properties signal for characterising speakers, especially formant dynamics; and secondly, speaker-distinguishing potential phonological due to sound change. In order explore these sources variability, large-scale database material collected simulated conditions being compiled. This paper outlines our motivation investigation, methods structure database, some findings change study.