Speech Therapy

A full-stack application for speech therapy assistance, combining a Python backend with a React TypeScript frontend.

Project Structure

speech-therapy/
├── backend/
│   ├── services/
│   │   ├── agent_orchestrator.py      # Orchestrates therapy agents
│   │   ├── mcp_server.py              # MCP (Model Context Protocol) server
│   │   ├── recover_audio.py           # Audio recovery utilities
│   │   └── repoint_jsonl.py           # JSONL data management
│   ├── deprecated_app_v1.py           # Legacy application (deprecated)
│   ├── sarvam_train.jsonl             # Training data for Sarvam
│   ├── whisper_train.jsonl            # Training data for Whisper
│   ├── whisper_train_clean.jsonl      # Cleaned Whisper training data
│   └── whisper_train_fixed.jsonl      # Fixed Whisper training data
│
├── frontend/
│   ├── src/                           # React/TypeScript source code
│   ├── public/                        # Static assets
│   ├── package.json                   # Frontend dependencies and scripts
│   ├── vite.config.ts                 # Vite build configuration
│   ├── tsconfig.json                  # TypeScript configuration
│   └── README.md                      # Frontend-specific documentation
│
└── .vscode/                           # VS Code settings

Technology Stack

Backend

• Language: Python
• Key Components:
  • agent_orchestrator.py - Agent coordination and orchestration
  • mcp_server.py - Model Context Protocol server implementation
  • recover_audio.py - Audio processing and recovery
  • repoint_jsonl.py - Data pipeline management

Frontend

• Framework: React 19.2.6
• Language: TypeScript (~6.0.2)
• Build Tool: Vite 8.0.12
• Development Tools: ESLint, TypeScript ESLint

Language Composition

• Python: 61.6%
• TypeScript: 22.4%
• CSS: 13.4%
• JavaScript: 1.6%
• HTML: 1.0%

Getting Started

Frontend Setup

cd frontend
npm install

Available Scripts

• npm run dev - Start development server
• npm run build - Build for production
• npm run lint - Run ESLint
• npm run preview - Preview production build

Backend Setup

The backend includes:

• Model training data in JSONL format for both Sarvam and Whisper models
• An MCP server for protocol-based communication
• Agent orchestration for therapy workflow coordination
• Audio processing utilities

Features

• Agent Orchestration: Coordinate multiple therapy agents
• Audio Processing: Recover and process audio data
• Model Training: Support for Sarvam and Whisper model fine-tuning
• MCP Server: Protocol-based communication interface
• React Frontend: Modern, responsive UI for therapy sessions

Development

This is a full-stack application with clear separation between frontend and backend services. The backend provides APIs and services that the frontend consumes through the MCP server interface.

License

Not specified

Contributing

Contributions are welcome! Please ensure code follows the project's ESLint configuration.

---

For frontend-specific details, see frontend/README.md