Pronouncing Rehabilitation of Hearing-Impaired Children Based on Chinese 3D Visual-Speech Database

Visual-speech thinking studied in the field of human-computer interaction is applied to voice assistant system. A new method based on 3D talking head driven by parameter is presented to make rehabilitation training and the speech rehabilitation problems of hearing-impaired children are resolved. Establish the 3D talking head model, and set up the 3D visual speech database by extracting parameters of the face, tongue and palate. By extracting parameters of speech database to drive 3D talking head, make pronunciation training for the deaf and no dumb children with hearing impairment, give the extraction method of face and tongue parameter, and introduce the specific implementation methods in details. Experiments show that the method can determine the difference position by comparing the standard speech database parameters and sequence parameters of trainers´ pronunciation to provide intuitionist information feedback for the trainees to improving the quality of their pronunciation.