The Digital India BHASHINI Division (DIBD) showed how multilingual AI can make research easier to access at ETD-2026, held at IIT Delhi. BHASHINI is exploring partnerships with higher education institutions and research repositories. The goal is to open up theses, dissertations and scholarly content across Indian languages.
Voice-First AI for Academic Settings
Ms. Baishali Chakraborty, Team Member, Ecosystem Enablement at BHASHINI, demonstrated voice-first multilingual AI, speech-to-text, translation, optical character recognition (OCR) and document translation. She showed how these tools can support multilingual access in academic and research settings.
A live demonstration of Shrutlekh drew attention. It transcribes speech in real time and translates it at the same moment. It could be used in lecture halls, research seminars, conferences and academic discussions.
BHASHINI by the Numbers
The session also covered BHASHINI’s wider ecosystem of language datasets, AI models, APIs and applications. BHASHINI has recorded over 10 billion cumulative AI inferences. It supports 36 Indian languages for text, 23 Indian languages for voice and 35 international languages. Its website translation plugin gives multilingual access across more than 800 government websites.
Translating Theses and Dissertations
A key focus was extending AI translation to electronic theses and dissertations (ETDs), including long, specialised academic documents. BHASHINI Pragati was presented as an example. It is a document translation workbench built for the Comptroller and Auditor General of India (CAG). It supports structured translation and review. It can process large documents of 100 to 500 pages, with quality checks and data security built in.
The discussion looked at adapting similar workflows for research repositories such as Shodhganga, maintained by INFLIBNET. Universities and other higher education institutions could use them too. Such tools could make theses, dissertations, research reports and other scholarly material available in Indian languages. They would keep document structure, specialised terms and meaning intact.
Harshad Sharma, Senior Manager (Finance) at DIBD, said making research accessible across Indian languages is an important step towards a more inclusive knowledge ecosystem. He said BHASHINI’s document translation and speech technologies can help universities, libraries and repositories reach wider audiences. Working with academic institutions will be key, he added. It will help BHASHINI understand their needs and build reliable solutions for multilingual access to research content.
Can AI Translate Specialised Research Reliably?
Participants asked about translating PDF documents on the fly. They also asked how reliable AI translation is for specialised academic and scientific content. Ms. Chakraborty stressed combining domain-specific machine translation with human review and quality assurance. The discussion also covered working with research organisations to build customised solutions for their repositories and academic fields.
More Tools in the Toolkit
The session covered BHASHINI’s OCR, which extracts text from scanned documents and images. It also covered terminology management through SHABD UDYAN. Transliteration helps preserve technical terms, names and expressions across scripts. Together, these tools can make research content easier to find and use, while respecting the language and formatting needs of academic documents.
BhashaDaan, BHASHINI’s citizen-contribution initiative, also came up. It builds language resources through speech and text contributions, audio transcription and text extraction from images and documents. Granthika was presented as a tool to digitize and preserve books, manuscripts, literary works, folk tales and oral histories.
An Open Invitation
BHASHINI invited higher education institutions, research repositories, libraries and academic organisations to collaborate. The work would involve choosing priority use cases, building domain-specific language resources and adapting translation workflows. This can address the needs of each repository, improve translation quality through human validation and widen access to India’s scholarly output.
BHASHINI’s presence at ETD-2026 reflects its wider aim of advancing language technology as a public digital resource. Working with academic institutions and repositories can carry multilingual AI into scholarly communication. It would help research and knowledge reach audiences across languages.
