MUMBAI, India, June 26 -- Intellectual Property India has published a patent application (202621052101 A) filed by Symbiosis International Deemed University on April 23, 2026, for A System And Method For Speech-Based Emotion Recognition In Legal Court Proceedings Using Deep Learning.
Inventors include Paneri Fulbandhe; Shreya Sonsare; Ankita Sorte; Dr. Monali Gulhane; and Dr. Nitin Rakesh.
The application for the patent was published on June 19, 2026, under issue no. 25/2026.
Abstract: ABSTRACT A SYSTEM AND METHOD FOR SPEECH-BASED EMOTION RECOGNITION IN LEGAL COURT PROCEEDINGS USING DEEP LEARNING The present invention relates to a system (100) and method for automated speech emotion recognition in legal court proceedings using deep learning. The system comprises an audio input module (110), a preprocessing module (120) for silence removal and audio standardization, a feature extraction module (130) implementing hybrid extraction combining Mel Frequency Cepstral Coefficients and Fast Fourier Transform features with emphasis on 300-3400 Hz frequency range, a data balancing module (140) utilizing Synthetic Minority Oversampling Technique, a classification module (150) comprising a Convolutional Neural Network with multiple convolutional layers, pooling layers, batch normalization, and dropout, and an output module (160) for emotion labeling. The system classifies speech into eight emotional categories including neutral, calm, happy, sad, angry, fearful, disgust, and surprised, providing objective emotional analysis to assist legal professionals in courtroom proceedings. [
Disclaimer: Curated by HT Syndication.
