Science TechnologySunday, 9 August 2026·The Hindu - Science
AI4Bharat at IIT Madras is building speech AI for India’s low-resource languages
AI4Bharat at IIT Madras is collecting speech data from 500+ districts to train AI for India’s diverse languages and dialects.
Key highlights
Direct fact
In August 2026, AI4Bharat at IIT Madras was highlighted for building speech AI that can understand and translate India’s low-resource languages, using data collected from more than 500 districts across the country.
Key specifics
- AI4Bharat is a research lab at IIT Madras.
- More than 500 districts were covered for speech-data collection.
- The project focuses on low-resource languages such as Tulu, Bundeli, Kodava and Santali.
- Researchers set up recording booths with local colleges and community organisations.
- Participants were encouraged to speak about Diwali, festivals, weddings and village life to capture accents and dialects.
Exam lens
Science and technology question type, AI4Bharat, IIT Madras, 500+ districts, low-resource languages; TNPSC may ask why Indian-language AI needs district-level speech data.