IIITH’s SARAL platform makes research simpler and more inclusive with AI-driven summaries

 To make scientific research more accessible and digestible to a broader audience, the International Institute of Information Technology, Hyderabad (IIITH), under the guidance of the Anusandhan National Research Foundation (ANRF), has developed SARAL. This AI-powered platform converts dense academic papers into concise, easy-to-understand video summaries in multiple Indian languages.

Built with a strong focus on accessibility, SARAL transforms LaTeX files, PDFs, or arXiv papers into short, 3-4 minute videos. It extracts key sections, Introduction, Methodology, Results, and Conclusion, and converts them into editable slides enriched with visuals. These slides are then turned into voice-backed video presentations in any of 11 supported Indian languages using Sarvam’s text-to-speech engine. The entire process leverages generative AI tools, including Gemini, Claude, and GPT.

The project was sparked by the growing need to simplify research communication beyond the academic circle. Prof. Ponnurangam Kumaraguru, leading the initiative, along with project leads and developers, aimed to reimagine how those without formal scientific training consume research. Inspired by tools like Google’s NotebookLM, which summarises complex documents into podcasts, SARAL focuses on video, a more engaging and widely shareable format.

The tool allows for significant customisation. Users can refine slide content, adjust visuals, and choose preferred audio voices and languages. While the current version produces static slides, efforts are underway to include animated video formats and even single-page research posters to suit academic conference needs.

SARAL is one of the early tools hosted on Democratiseresearch. in a platform envisioned to make both academic and industry research discoverable and understandable to a wider audience. The team is also exploring similar tools for simplifying patents, aiming to close the gap between researchers and industry stakeholders and promote technology transfer.

The source code for SARAL is publicly available on GitHub, and IIITH plans to host hackathons and competitions to drive further innovation around research communication. As AI continues to transform academic processes, SARAL offers a much-needed bridge between science and society, making the pursuit of knowledge a bit less elusive and a lot more inclusive.