MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202641090847 A) filed by Dr. M. M. Gopi; Dr. A. Madhusudhana Rao; Dr. S Bhanu; Dr. D. V. Adithya Joseph; Dr. S. Padmaja; Dr. K. Madhu Murthy; Saidabhanu Shaik; T Harika; and K K Hari on July 27, 2026, for System And Method For Real-Time Prosodic Feature Extraction And Learner-Adaptive Pronunciation Error Classification Impaired Users.

Inventors include Dr. M. M. Gopi; Dr. A. Madhusudhana Rao; Dr. S Bhanu; Dr. D. V. Adithya Joseph; Dr. S. Padmaja; Dr. K. Madhu Murthy; Saidabhanu Shaik; T Harika; K K Hari; and Dr. S. Kannappan.

The application for the patent was published on July 31, 2026, under issue no. 31/2026.

Abstract: A computer-implemented system (100) and method for real-time prosodic feature extraction and learner-adaptive pronunciation error classification are disclosed. The system comprises a central processing unit (110), an audio acquisition module (120), a prosodic feature extraction engine (130) that concurrently extracts pitch contour, formant trajectories, stress and duration measures, and speech rate from a learner's utterance to form a composite prosodic feature vector, a learner profile module (140) that determines a native-language interference profile and retrieves a corresponding interference-pattern data structure (142), an adaptive error classification engine (150) comprising an adaptive weighting module (151) that computes feature-weighting coefficients as a function of the interference-pattern data structure and an error- type classifier and severity scorer (152, 153) operating on the resulting weighted feature vector, and a feedback generation module (160) that generates real-time corrective feedback and targeted exercise recommendations, with a learner progress storage (180) refining the interference profile over successive sessions. The invention provides a closed-loop, interference-profile-conditioned architecture for real-time, learner-specific pronunciation error detection and correction.

Disclaimer: Curated by HT Syndication.