MUMBAI, India, Oct. 26 -- Intellectual Property India has published a patent application (202641123016 A) filed by Mlr Institute Of Technology; and Marri Laxman Reddy Institute Of Technology And Management on October 14, 2026, for Adaptive Multi-Layer Real-Time Speaker Recognition System With Dynamic Acoustic Fingerprinting And Context-Aware Identity Verification.
Inventors include Dr. G. V. Subba Reddy; Dr. B. Kiran Kumar; Ms. B. Apurupa; and Mr. B. Varun.
The application for the patent was published on October 23, 2026, under issue no. 43/2026.
Abstract: The present invention relates to a computer-implemented system for real-time speaker recognition and identity verification using dynamic acoustic fingerprinting. The system comprises an audio acquisition module configured to receive live audio streams; a preprocessing module configured to perform noise suppression, signal normalization, and compensation for variations in microphone characteristics, channel distortion, and reverberation; and a multi-layer acoustic fingerprint extraction module configured to derive speaker-specific features comprising resonance characteristics, harmonic distribution, pitch contours, breath-related features, articulation characteristics, and speech rhythm. A behavioural speech analysis module extracts additional features comprising speaking rate, pause patterns, pronunciation characteristics, and habitual lexical patterns. A feature fusion and identity verification module combines the extracted acoustic and behavioural features to generate speaker representations, calculate similarity scores, and determine identity verification results based on predefined or adaptively updated decision thresholds. A spoof detection module analyses audio characteristics to identify potential replay attacks, synthetic speech, cloned voices, and manipulated audio recordings. A confidence- weighted adaptation module updates speaker representations using selected audio samples subject to confidence and identity-consistency criteria. The system generates speaker identification results, verification confidence scores, and audio authenticity assessments for real-time applications. The invention integrates adaptive acoustic processing, multi-layer feature extraction, speaker representation, identity verification, and spoof detection within a unified computational architecture for voice-based authentication, communication systems, and human-machine interaction platforms. 5 Claims & 3 Figure.
Disclaimer: Curated by HT Syndication.
