MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641080412 A) filed by Cmr Institute Of Technology on June 30, 2026, for Web-Based Artificial Intelligence-Assisted Speaking Confidence Practice Platform With Real-Time Multimodal Behavioral Analysis And Adaptive Scoring System.
Inventors include Pavan Kumar Panakanti; Sri Rama Lakshmi Reddy; K. Ashitosh Reddy; P Varshit; K. Srividya; and K. Nandini.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: A web-based artificial intelligence-assisted speaking confidence practice platform is disclosed, comprising an integrated system of real-time multimodal behavioral analysis components implemented as a browser-executable application. The platform enables users to engage in structured speaking practice sessions in which their verbal delivery behaviors are simultaneously analyzed through a real- time audio analysis subsystem that employs the browser-native Web Audio API to compute speech continuity ratios, pause frequency counts, and silence duration metrics, and a real-time video analysis subsystem that employs the MediaPipe facial landmark detection machine learning library to compute eye contact maintenance ratios and facial expressiveness indicators from per-frame analysis of the user's camera video stream. A behavioral confidence scoring engine synthesizes the computed audio-derived and video-derived behavioral metrics through a weighted composite scoring formula to produce a normalized, session-level behavioral confidence score accompanied by a categorical performance tier classification. An artificial intelligence feedback generation subsystem constructs a structured prompt payload from the session metrics and transmits it to a large language model application programming interface to produce a concise, personalized, actionable feedback narrative that is presented to the user upon session completion. A cloud-based infrastructure comprising Firebase Authentication for secure token-based user identity management and a hierarchically structured cloud database for session record persistence enables users to retrieve their complete speaking session history through an interactive analytics dashboard that renders time-series visualizations of their longitudinal confidence score trajectories. The platform requires no specialized software installation and operates entirely within standard web browser environments, providing universal accessibility for students, job aspirants, and professionals seeking objective, behavior-based improvement of their speaking confidence.
Disclaimer: Curated by HT Syndication.
