How fatma-moanes/voice-assistant works
Main Components
A. Frontend Layer:
• Streamlit UI Components (from app/components/)
• Chat Display
• Controls
• Sidebar
• Custom CSS
B. Core Application Layer:
• Configuration Management
• Database Interface
• Agent System
C. Service Layer:
• Speech-to-Text Services
• Text-to-Speech Services
• LLM Integration
D. External Services:
• MongoDB
• AWS Services (Polly, Transcribe)
• LLM Providers (GROQ/OpenAI)
• Whisper Streaming
[User Interface Layer]
• ↕️
[Core Application Layer]
• ↕️
[Service Layer]
• ↕️
[External Services Layer]
Component Relationships
• User Interface → Speech Services (audio input/output)
• Speech Services → External APIs (AWS, Whisper)
• Core Agent → LLM Providers
• Core Agent → Database
• Service Layer → External Services
Detailed Component Breakdown
Frontend Layer:
[Chat Display] ← → [Controls]
• ↕️
[Sidebar] ← → [Audio Controls]
Core Layer:
[Config Manager] → [Agent System]
• ↓
[Database Interface]
Service Layer:
[STT Service] → [Agent Tools]
[TTS Service] ← [Chatbot]
Additional Elements
Supporting components:
• Configuration files (config.yml)
• Environment variables
• Logging system
• Debug mode components
Technical Details to Highlight
• MongoDB integration for appointment data
• Multiple STT options (Whisper, AWS, Deepgram)
• LLM provider flexibility (GROQ/OpenAI)
• Bilingual support architecture
• Real-time audio processing pipeline
The separation of concerns between UI, core logic, and services
This system design should clearly communicate how the voice assistant processes user input, manages conversations, and integrates with various services while maintaining a clean separation of concerns.