MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202621064112 A) filed by Prof. Shobha Mahendrakumar Patil; Aastha Nainabhiram Tiwari; Amogh Mahesh Yadav; Ankit Sachin Vadnere; and Snehal Surendra Thawase on May 21, 2026, for A Multilingual Ai Framework For Real-Time Semantic Simplification Of Spoken Communication..
Inventors include Prof. Shobha Mahendrakumar Patil; Aastha Nainabhiram Tiwari; Amogh Mahesh Yadav; Ankit Sachin Vadnere; and Snehal Surendra Thawase.
The application for the patent was published on August 07, 2026, under issue no. 32/2026.
Abstract: The present invention provides a computer-implemented multilingual accessibility engine that semantically transforms complex spoken and written content into simple, comprehensible regional language output for Deaf, Hard-of-Hearing, and low-literacy users. The system integrates real-time audio transcription, Large Language Model based semantic simplification, Code Mixing, and multilingual delivery across three environments — Telegram bot, Zoom live meeting, and physical classroom — within a single unified pipeline powered by Gemini 2.5 Flash. A 60-Second Pulse summarization mechanism prevents cognitive overload during live sessions. Simplified output is delivered via Telegram chat reply or a real-time SSE-based web application, supporting regional Indian languages including Hindi, Marathi, Bengali, and Arabic, making it a viable accessibility solution for DHH and low-literacy populations.
Disclaimer: Curated by HT Syndication.
