MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202641075788 A) filed by Vardhaman College Of Engineering on June 18, 2026, for Method And Apparatus For Biometric Voice Extraction And Noise Suppression.
Inventors include K. Sripal Reddy; S. Srinivas; Ch. Sulakshana; Ch. Sreenivasa Rao; Dr. V. Purna Chandra Reddy; and K. Narsimha Reddy.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: Method and Apparatus for Biometric Voice Extraction and Noise Suppression is the proposed invention. The invention proposed herein includes capturing audio signals using one or more microphones and processing the signals utilising deep learning-assisted acoustic analysis, transformer-based temporal modelling, and intelligent signal enhancement techniques. It includes a biometric voice extraction engine that analyses pitch patterns, harmonic structures, vocal tract resonance, and speech dynamics unique to the speaker to create voice embeddings for secure authentication. At the same time, an adaptive noise suppression module employs beamforming, spectral masking, source separation, and frequency domain filtering to suppress environmental noise, reverberation, and overlapping speech, while preserving the essential biometric voice features. The spatial localisation algorithms estimate the direction of the target speakers to enable selective speech enhancement in multi- speaker scenarios. The architecture is optimised for deployment in smart communication systems, security platforms, hearing assistance devices, automotive systems and voice-controlled applications by leveraging lightweight neural network quantisation and low-latency edge AI processing. The proposed invention provides improved speech intelligibility, improved biometric extraction accuracy, adaptive environmental robustness and reliable real-time speaker authentication under noisy and dynamically changing conditions.
Disclaimer: Curated by HT Syndication.
