An Emotion Aware AI-Based Virtual Therapy System Using Text and Speech Analysis

Mental health support systems based on conventional chatbots often fail to provide emotionally appropriate responses due to their limited understanding of user emotions. To address these limitations, this paper presents an emotion-aware virtual therapy system that utilizes both textual and speech input to detect the emotional state of users and generate context sensitive therapeutic responses. The proposed system integrates natural language processing techniques for text-based emotion recognition and deep learning models for speech emotion analysis using acoustic features such as melfrequency cepstral coefficients. A multimodal architecture is designed to improve overall accuracy and reliability. The system is implemented as a mobile application to ensure accessibility and real time interaction. Experimental evaluation conducted on benchmark emotion datasets demonstrates that the proposed approach achieves higher performance compared to unimodal emotion detection methods. The results indicate that incorporating emotion awareness significantly enhances the effectiveness of virtual therapy systems by enabling personalized and empathetic user interactions. This work highlights the potential of multimodal artificial intelligence in scalable and intelligent mental health support solutions.