An advanced, multi-modal AI system designed to offer empathetic mental health support through real-time text and voice interaction. Utilizes a fine-tuned BlenderBot-400 M-distill model with Low-Rank Adaptation (LoRA) for optimized conversational response. The system is built on a React + TypeScript frontend, powered by a Flask RESTful API backend, and integrates Azure Speech Services for speech-to-text and text-to-speech capabilities. Designed with a privacy-first architecture and optimized for low-latency, human-like dialogue delivery.