Silent Supporter - AI Therapy Platform
Scalable multimodal therapy platform with advanced AI capabilities
Project Overview
Silent Supporter is an innovative AI-powered therapy platform developed in collaboration with Tsinghua AIR, featuring advanced multimodal capabilities including real-time emotion visualization and personalized music therapy.
Key Achievements
- Performance Optimization: Reduced real-time generation latency by 80% through WebSocket implementation
- AI Model Enhancement: Achieved 3x inference speed improvement via ONNX Runtime graph optimization
- Custom AI Training: Fine-tuned InspireMusic 1.5B model for targeted music style transfer
- Real-time Visualization: Developed WebGL engine for dynamic conversational emotion display
Technical Architecture
Backend Infrastructure
- Node.js: High-performance JavaScript runtime for scalable backend services
- PostgreSQL: Robust relational database for primary data storage
- Redis: In-memory caching for session management and performance optimization
- RabbitMQ: Event-driven architecture for asynchronous multimodal generation
AI & Machine Learning
- Custom Model Fine-tuning: InspireMusic 1.5B model adapted for therapeutic music generation
- ONNX Runtime: Graph optimization for 3x faster inference speeds
- Multimodal Processing: Real-time audio, text, and visual data integration
Frontend & Visualization
- WebGL Engine: Custom-built real-time emotion visualization system
- Real-time Updates: WebSocket-powered live emotional state tracking
- Responsive Design: Cross-platform compatibility for therapy sessions
Innovation Highlights
This project represents a significant advancement in digital therapy platforms, combining cutting-edge AI research with practical therapeutic applications. The integration of real-time emotion visualization with personalized music therapy creates a uniquely immersive therapeutic experience.
Research Collaboration
Developed in partnership with Tsinghua AIR, this project contributes to ongoing research in AI-assisted mental health interventions and demonstrates the practical application of advanced machine learning techniques in healthcare technology.