For example,Бобцов

BILINGUAL MULTIMODAL SYSTEM FOR TEXT-TO-AUDIOVISUAL SPEECH AND SIGN LANGUAGE SYNTHESIS

Annotation

We present a conceptual model, architecture and software of a multimodal system for audio-visual speech and sign language synthesis by the input text. The main components of the developed multimodal synthesis system (signing avatar) are: automatic text processor for input text analysis; simulation 3D model of human's head; computer text-to-speech synthesizer; a system for audio-visual speech synthesis; simulation 3D model of human’s hands and upper body; multimodal user interface integrating all the components for generation of audio, visual and signed speech. The proposed system performs automatic translation of input textual information into speech (audio information) and gestures (video information), information fusion and its output in the form of multimedia information. A user can input any grammatically correct text in Russian or Czech languages to the system; it is analyzed by the text processor to detect sentences, words and characters. Then this textual information is converted into symbols of the sign language notation. We apply international «Hamburg Notation System» - HamNoSys, which describes the main differential features of each manual sign: hand shape, hand orientation, place and type of movement. On their basis the 3D signing avatar displays the elements of the sign language. The virtual 3D model of human’s head and upper body has been created using VRML virtual reality modeling language, and it is controlled by the software based on OpenGL graphical library. The developed multimodal synthesis system is a universal one since it is oriented for both regular users and disabled people (in particular, for the hard-of-hearing and visually impaired), and it serves for multimedia output (by audio and visual modalities) of input textual information.

Читать текст статьи

Keywords

Articles in current issue