MUMBAI, India, Jan. 23 -- Intellectual Property India has published a patent application (202541116147 A) filed by Nandha Engineering College, Erode, Tamil Nadu, on Nov. 24, 2025, for 'real-time multi user speech-to-speech ai platform.'

Inventor(s) include T Rajasekaran; S Balahariharan; M Kavikumar; S S Arunesh; M S Phurnes; J A Bharathkumar; and M Fathima Maahira.

The application for the patent was published on Jan. 23, under issue no. 04/2026.

According to the abstract released by the Intellectual Property India: "The Real-Time Speech-to-Speech AI Communication Systeni introduces a multi-user, Al-driven conversational platform designed to enable natural voice interaction between an artificial intelligence agent and multiple participants within a shared virtual room. The invention integrates a Low-Latency Audio Stream Processing Engine built using LiveKit protocols to capture, synchronize, and transport speech from all users in real time. This audio stream is continuously ingested and processed by an advanced Speech-to-Text module, enabling the system to interpret speech with high accuracy. The Multi-User Conversational Intelligence Core, powered by a large language model, analyzes contextual intent, resolves overlapping speech inputs, and generates meaningful, coherent responses that maintain conversation flow across diverse group settings. The Dynamic Voice Synthesis & Response Delivery Module converts the AI-generated text into natural-sounding speech and streams it back instantly to all participants through the LiveKit framework: This coordinated processing pipeline ensures minimal latency, high scalability, and seamless communication, allowing the AI agertt tO function as art active virtual participant in any meeting, classroom, or collaborative environment. The system offers significant advantages, including reduced operational effort, self-hosting capability, easy integration with modern web applications, and an enhanced communication experience that brings human-like intelligence into real-time multi-user conversations."

Disclaimer: Curated by HT Syndication.