Viseme set and ASCII mapping codes preparation for standard Malay audiovisual speech synthesis / Siti Salwa Salleh … [et al.]

This paper presents the preparation aspects of identifying ASCII mapping code for Standard Malay (SM) Phonetics alphabet and visemes set in developing a Standard Malay (SM) audiovisual speech synthesis (AVSS). We describe our effort on the examination of existing visemes set which can be used in presenting SM mouth shapes during speech event. The results of this analysis will be employ in building up a talking head, a computer generated model of speaking faces which integrate a speech synthesis module to the visual representation of facial animation The visemes and phonemes will be coupled tightly in the AVSS to provide naturalness of the speaking talking head. The study starts by identifying the SM phonemes and dissect the similarity and differences between SM and American-English phoneme. The purpose is to identify the additional phonemes present in SM but not in the American-English. The result would lead to the establishment on the viability to use and extend the DECTAlk visemes set in SM audiovisual speech synthesis. To obtain the SM visemes set, a video recording were done. The video recording captures an event of a subject who pronounced a collection of words which consist of every phoneme in SM. In addition to allow the manipulation of the phonetic alphabet in the ASCII computer system, our effort also proposed the setting up of SM phonetics alphabet mapping codes. The mapping procedure followed the guidelines which suggested by the Speech Assessment Methods (SAM) community.