Collaboration
ITRI and PTS Create AI Sign Language Weather Broadcaster for the Deaf Community
ITRI and Taiwan’s Public Television Service (PTS) announced a joint project on developing an AI sign language weather broadcaster, with plans for the service to go live as early as the second half of 2025. The partnership aims to increase the accessibility of weather information and enable deaf individuals to receive the latest warnings for natural disasters and other emergency incidents.
ITRI Chairman Tsung-Tsong Wu (right) and Taiwan Public Television Service (PTS) Foundation Chairperson Yuan-Hui Hu (left) announced the collaboration on developing the AI sign language broadcaster.
“Technology arises from human needs, and AI can play a key role in helping those in need,” said ITRI Chairman Tsung-Tsong Wu. “The collaboration between ITRI and PTS is a demonstration of inclusive technology that offers everyone equal opportunity to enjoy the benefits of AI. The AI sign language weather broadcaster is expected to compensate for the shortage of sign language interpreters and provide the deaf equitable access to information via mass media.”
Taiwan Broadcasting System and Taiwan Public Television Service (PTS) Foundation Chairperson Yuan-Hui Hu remarked that the PTS has been dedicated to producing sign language TV programs for the Deaf community and is actively leveraging technology to make the media environment more accessible to them. “Information equality is a fundamental value in modern society,” stated Hu. He hopes to help deaf individuals stay up to date during typhoons, earthquakes, and other emergencies through advanced communication technology.
The AI sign language weather broadcaster accurately captures complex hand movements to create smooth and realistic 2D animations.
ITRI Vice President and General Director of Information and Communications Research Laboratories Pang-An Ting remarked that ITRI began developing the AI sign language interpreter and building Taiwan’s sign language corpus in 2023. “Sign language is a visual language that conveys meaning through a combination of facial expressions and hand gestures. An AI sign language broadcaster requires high-precision skeleton estimation technology and deep learning gesture generation models to accurately present hand movements. This is much more complicated than current AI audio and video broadcasting, which deals with speech generation and lip-syncing technologies,” he explained.
Dr. Ting further pointed out that sign language evolves with variations in local dialects, emphasizing the need to build a sign language corpus in Taiwan. “We need both professional sign language interpreters and members from the Deaf community to work together to build the corpus. We hope that this collaborative project will promote information equality and also preserve Taiwan’s sign language as an important cultural asset,” he said.