Speech-to-Thai Sign Language Conversion for Thai Deaf: A Case Study of Crime News

This paper aims to propose speech-to-Thai sign language conversion in crime news by using a crime news video as the input and a sequence of Thai sign gesture videos as the output. There are three main modules. Firstly, Speech-to-text module is to convert speech in crime news video to text. Next, the Thai words Segmentation module is used to tokenize the paragraph of text to the sequence of words. Lastly, Gestures sequence construction module is applied form the sequence of gesture videos. Also, this research has been cooperated in evaluating results from the National Association of the Deaf in Thailand in terms of the benefits of this research and suggestions for future improvements.