Synthia - AI-Powered Healthcare Assistant

Overview

Synthia is a real-time voice AI assistant that combines OpenAI's Whisper for speech-to-text, ElevenLabs for text-to-speech, and GPT-4 for natural conversation. The platform delivers an empathetic and intelligent virtual assistant specifically trained for healthcare scenarios, with a focus on dementia care support.

Demo

Chat Interface

Video Playthrough

Watch Demo Video

Technical Architecture

AI & Voice Processing Stack

• Language Model: OpenAI GPT-4o-mini for contextual understanding and natural conversation
• Speech-to-Text: OpenAI Whisper API for accurate voice transcription
• Text-to-Speech: ElevenLabs API for natural, human-like voice synthesis
• Audio Processing: Web Audio API for real-time audio handling and playback

Core AI Features

• Real-time voice recognition using OpenAI Whisper
• Natural voice synthesis powered by ElevenLabs
• Contextual AI responses using OpenAI GPT-4
• Seamless voice conversation pipeline
• Healthcare-specific conversation context

Project Structure

src/
├── services/      # AI and voice processing integrations
│   ├── ai-services.ts    # OpenAI GPT-4 integration
│   ├── speech-to-text.ts # OpenAI Whisper integration
│   └── text-to-speech.ts # ElevenLabs integration
└── types/         # TypeScript type definitions

Next Steps

Immediate Priorities

1. Implement comprehensive error handling for AI service failures
2. Expand the healthcare knowledge base with domain-specific training
3. Optimize voice processing latency
4. Implement conversation history persistence
5. Add fallback mechanisms for API outages

Future Enhancements

1. Multi-language support with Whisper's multilingual capabilities
2. Custom voice model training with ElevenLabs
3. Fine-tuned GPT models for healthcare-specific responses
4. Integration with healthcare APIs for real-time data