15th International Congress of Phonetic Sciences (ICPhS-15)

Barcelona, Spain
August 3-9, 2003


Gender- and Speaker-Independent Prediction of Vocalic Lipshapes Using Acoustic Cues

Michiko Inoue

University of Munich, Germany

In this paper we describe the results of perceptual experiments with static lip configurations (visemes) and the synthesis of (expected) vocalic lip postures on the basis of F0 and formant measurements. Three steps were performed: (i) two visual perception experiments of vocalic lip photographs, (ii) a graphic experiment using cardinal vowel lip configurations, and (iii) a lip configuration estimation through a two-dimensional linear regression model and a manual optimization. We assume a linear relation between lip configuration and acoustics.

Full Paper

Bibliographic reference.  Inoue, Michiko (2003): "Gender- and speaker-independent prediction of vocalic lipshapes using acoustic cues", In ICPhS-15, 3209-3212.