Talk in ISL: Empowering the Deaf Community in India
Bridging the gap
India is home to nearly 63 million deaf individuals. Alarmingly, only about 26% of the 15 million deaf individuals of working age are employed. This low employment rate can be attributed to two main factors: a lack of education within the deaf community and the inherent challenges they face in effective communication. Providing access to technology that helps them engage with society, be more productive, and realise their potential could significantly improve their quality of life.
Research in sign languages
Sign languages — American Sign Language (ASL), Chinese Sign Language (CSL), British Sign Language (BSL), and others — have been studied for decades and are now widely accepted as bona fide natural languages. Being a visual medium, sign language lends itself well to machine-learning-based solutions. Various research organisations, universities, and startups are working in education and communication using sign languages.
Efforts in India
In India, the Indian Sign Language Research and Training Centre (ISLRTC) focuses on research and training in Indian Sign Language (ISL). The Noida Deaf Society, an NGO we partner with, works on ISL training and on making deaf individuals economically productive. Several other organisations are producing content in ISL to benefit the deaf community.
The Talk in ISL (TIISL) project
One of the pioneering projects by the Longtail AI Foundation (LAF) is to address the communication and education challenges faced by the deaf community in India. The Talk in ISL (TIISL) project aims to facilitate meaningful communication between deaf and hearing individuals.
Data is the cornerstone of machine-learning-based solutions. Amassing a substantial amount of ISL content is therefore crucial for training models and making their output reliable. The immediate goal of the TIISL project is to gather a comprehensive ISL dataset. We are also developing technology to generate synthetic data to augment this dataset. Achieving a large dataset would mark a significant milestone in ML-based ISL research, positioning it alongside mature datasets like BOBSL, CSL-Daily, and Phoenix14T.
Core principles
Openness. The ISL dataset and the technology built upon it will be accessible to researchers and to individuals who can use it to improve their daily lives and opportunities.
Responsibility. The technology developed will responsibly facilitate communication between deaf and hearing individuals. This responsibility requires ongoing improvements in model development.
Quality. Being open and free does not compromise quality. On the contrary, the commitment to responsibility requires the service quality match that of any paid alternative.
By focusing on these principles, the TIISL project aims to revolutionise communication and education for the deaf community in India, fostering inclusivity and empowerment.
We are always seeking ML researchers to join us. Write to info@longtailai.org.