Coarticulation Analysis for Sign Language Synthesis

Lucie Naert 1 Caroline Larboulette 1 Sylvie Gibet 1
1 EXPRESSION - Expressiveness in Human Centered Data/Media
UBS - Université de Bretagne Sud, IRISA-D6 - MEDIA ET INTERACTIONS
Abstract : A sign language utterance can be seen as a continuous stream of motion, involving the signs themselves and inter-sign movements or transitions. Like in speech, coarticulation constitutes an important part of the language. Indeed, the signs are contextualized: their form and, most of all, the transitions will greatly depend on the surrounding signs. For that reason, the manual segmentation of sign language utterances is a difficult and imprecise task. Besides, annotators often assume that both hands are synchronous, which is not always true in practice. In this paper, we first propose a technique to automatically refine the segmentation by adjusting the manual tags isolating signs from transitions. We then study motion transitions between consecutive signs and, in particular, the duration of those transitions. We propose several computation techniques for the transition duration based on the analysis we have conducted. Finally, we use our findings in our motion synthesis platform to create new utterances in French Sign Language.
Complete list of metadatas

Cited literature [26 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01649815
Contributor : Caroline Larboulette <>
Submitted on : Monday, November 27, 2017 - 6:19:50 PM
Last modification on : Thursday, February 7, 2019 - 3:06:19 PM

File

uahci2017_naert_cr_withpagenum...
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01649815, version 1

Citation

Lucie Naert, Caroline Larboulette, Sylvie Gibet. Coarticulation Analysis for Sign Language Synthesis. International Conference on Universal Access in Human-Computer Interaction, Jul 2017, Vancouver, Canada. ⟨hal-01649815⟩

Share

Metrics

Record views

359

Files downloads

195