Search

This paper presents a method of tracking points on a speaking face and the reconstruction of a face model from speaking descriptive vectors followed. After capturing a video of a talking face using a CCD camera, his facial expression changes in a video sequences. The essencial of this work is to track the feature points on the face using the Lucas-Kanade method, then use descriptive vectors up to be exploited for the reconstruction of a model of speaking face. The tracking method used is based on the theorem of Lucas-Kanade and facial reconstruction based on the method of delaunay triangulation.

We acoustically analyzed behavior of speech signal produced in noisy constraint by four speakers, when noise is sent by a helmet to speaker and when noise is sent by high speaker. The goal is to find speech signal acoustic parameters which are most sensitive to noise and the compensatory strategies adopted by the speakers to counter this constraint. Results obtained in both cases show a great constraint influence over timing, energy, fundamental frequency F0, formants (F1, F2) of the analyzed units. They can be summarized such as: (1) In temporal field: We noted an increase in analyzed units duration. We can conclude that speaker tends to slow down his elocution rhythm with constraint in order to better render comprehensible itself; (2) In energy field: there is an increase in intensity. Thus the speaker speaks extremely in noise constraint; (3) In frequential field: (a) We noted an increase in formants F0 and F1 but F2 stay stable in spite of noise constraint. Its meaning that the speakers, while speaking extremely, increase their aperture but the vowels anteriority does not change. (b) Vocalic spaces are reduced with centralization. Its meaning that the speakers modify their formants to avoid constraint while remaining understandable.

We studied in this work the influence of labial constraint in speech production. We acoustically analyzed vowels behavior in constraint, then the constraint influence on the coarticulation on 4 speakers. Results obtained show a great influence of the constraint on the acoustic parameters in particular on the frequential parameters which are key parameters in synthesis, recognition and coding systems of the speech and in orthophonics reeducation programmes for handicapped persons. These results can be summarized such as: In the temporal field: A temporal reduction in the analyzed units, therefore the speaker tends to accelerate his rate/rhythm of elocution with the constraint. In the frequential field: (1) Spaces vocalic F2 (F1) are reduced but without centralization of the vowels. Thus the speakers reach with difficulty their acoustic targets with the constraint to see not at all; (2) there is variation of fundamental frequency according to the formants F0 (F1), F0 (F2) and F0 (F3). However direction of variation depends on speaker. Thus constraint varies the intonation; (3) locus equation slope obtained increase all with constraint. Thus labial constraint increases coarticulation degree.