Research Scope and Objectives
The project focuses on developing Thai Sign Language intelligence through five integrated work packages: Sign2Text translation from sign-language video to Thai text, Text2Sign translation from Thai text to sign-language video, sign-language avatar development, intelligent sign-language API infrastructure, and practical applications such as a mobile translation app, avatar-based interpretation for Thai PBS online documentary videos, avatar support for online learning, and dissemination activities. The objective is to improve the quality of life of people with hearing disabilities by enabling better access to information on television, the internet, education platforms, and everyday communication contexts.
Donor, Partners, and Geographical Scope
The project is funded by the Broadcasting and Telecommunications Research and Development Fund for Public Interest under the Office of the National Broadcasting and Telecommunications Commission (NBTC), with a total budget of approximately THB 13.3 million and a 36-month implementation period from 17 December 2025 to 16 December 2028. The project is led by the Asian Institute of Technology (AIT), with collaboration and stakeholder engagement involving sign-language interpreters, Thai Sign Language experts, linguists, the Deaf Association of Thailand, representatives of deaf and hard-of-hearing communities, and public media use cases such as Thai PBS. The geographical scope is Thailand, with emphasis on Thai Sign Language and public-service accessibility for Thai deaf and hard-of-hearing users.
AIT Role and Responsibilities
AIT serves as the lead implementing institution and is responsible for project management, research design, technical development, dataset construction, model training and evaluation, studio design, ethics preparation, stakeholder coordination, and dissemination. AIT’s research team will coordinate the development of Thai Sign Language datasets, AI models for Sign2Text and Text2Sign, avatar generation, API infrastructure, and prototype applications, while working with sign-language experts, linguists, interpreters, and deaf-community representatives to ensure linguistic correctness, usability, and social relevance.
Expected outcomes and impact
The project is expected to deliver a mobile sign-language avatar application, a sign-language avatar model for online video media, source code on GitHub, a Thai Sign Language AI model, intelligent sign-language API infrastructure, an open Thai Sign Language dataset, at least 10 educational or public-information programs with avatar interpretation, online learning materials with sign-language avatar support, a sign-language recording studio, one knowledge dissemination seminar with at least 100 participants, and three international journal submissions. The intended impact is to expand communication access for people with hearing disabilities, reduce barriers in media and education, support future research on Thai Sign Language processing, and provide a foundation for national-level accessibility services
Lessons learned and way forward
The early phase of the project shows that high-quality Thai Sign Language technology cannot rely only on AI model development; it requires a standardized corpus, careful annotation, linguistic review, controlled video-recording procedures, ethical approval, and continuous participation from deaf users and sign-language experts. Existing Thai Sign Language resources are still fragmented, often dictionary-based or small-scale, so the next phase should prioritize systematic vocabulary and content selection, video recording with quality control, gloss and Thai-text annotation, stakeholder validation, model training, avatar evaluation, API design, and user testing. The way forward is to strengthen community participation, ensure ethical and inclusive data collection, and progressively transform the research outputs into usable public services for media, education, and daily communication.
