Page 741 - AI for Good Innovate for Impact
P. 741
AI for Good Innovate for Impact
Use Case 3: From Silence to Speech: A Novel AI System for Speech
Development for the Deaf Through Lip-Reading 4.9: Accessibility
Organization: Huazhong University of Science and Technology
Country: China
Contact Person(s):
Ran Wang, rex _wang@ hust .edu .cn
Yang Xiao, Yang _Xiao@ hust .edu .cn
1 Use Case Summary Table
Item Details
Category Accessibility
Problem Addressed By leveraging Artificial Intelligence(AI) technologies to enable individuals
with hearing impairments to regain vocal communication capabilities, this
initiative aims to mitigate educational resource inequities stemming from
the scarcity of human and equipment resources.
Key Aspects of Solu- Develop an AI-powered lip-reading recognition system that converts
tion visual facial signals into textual language information through system
processing. This system, integrated with standardized video materials, will
serve as a self-learning guide for deaf individuals to study lip language.
The platform will be disseminated through online networks to ensure
enough educational resources.
Technology Deep Learning, Vision-language model
Keywords
Data Availability Public Data source of Lip Reading: CAS-VSR-W1k [1]
Metadata (Type of Videos
Data)
Model Training and Our vision-language model was trained and tested on public video data-
Fine-Tuning sets.
Testbeds or Pilot The AI algorithms are deployed on cloud-based servers, and the product
Deployments is delivered to deaf users through client-side applications.
Link: [2]
705
