MUMBAI, India, Feb. 6 -- Intellectual Property India has published a patent application (202541087094 A) filed by Dr. Mamatha S K; Sreya K J; RNS Institute Of Technology; and Rashmi B C, Bangalore, Karnataka, on Sept. 12, 2025, for 'multimodal emotion classification using bert and wav2vec: a fusion of text and audio modalities.'
Inventor(s) include Dr. Mamatha S K; Rashmi B C; Sreya K J; and Richa Sinha.
The application for the patent was published on Feb. 6, under issue no. 06/2026.
According to the abstract released by the Intellectual Property India: "Emotion recognition has garnered significant interest in recent years due to its profound implications for enhancing human-computer interaction (HCI) systems. By enabling machines to detect, interpret, and respond to human emotions, it opens avenues across various applications, including virtual assistants, mental health monitoring, e-learning platforms, and affective computing interfaces. The ability to perceive emotions accurately enhances the contextual understanding of user intent, thereby leading to more natural, empathetic, and effective interaction paradigms. Here, we present a robust multimodal deep learning framework for emotion recognition by integrating textual and acoustic modalities. Textual features are extracted using BERT embeddings, while Wav2Vec2 representations are employed for acoustic features. The embeddings are concatenated and fed into a fully connected neural network for classification. Experiments conducted on the MELD dataset, highlighting the complementary nature of text and speech in emotional expression. Additionally, waveform visualizations for different emotions offer intuitive insights into acoustic patterns. This work contributes toward building more emotionally intelligent systems."
Disclaimer: Curated by HT Syndication.
