Education Technology Project

A comprehensive education technology platform featuring a modern frontend UI and an advanced Text-to-Speech (TTS) generation system powered by MoonCast.

🏗️ Project Overview

This project consists of two main components:

1. Frontend UI - A modern web interface for user interaction
2. TTS Generation System - Advanced audio generation using MoonCast and GPT-4.1

Current Status

⚠️ Note: The Docker container integration with RunPod is currently not functional, so the frontend cannot generate TTS directly. However, the frontend serves as a complete UI design showcase, and the TTS system works independently.

🎯 TTS System Features

The TTS system under MoonDIA/trained_mapper provides:

• Multi-speaker audio generation with consistent voice characteristics
• GPT-4.1 integration for intelligent script generation
• Scalable local processing with sliding window optimization
• Semantic token conversion for high-quality audio output
• Configurable duration (currently limited to 15,000 tokens, expandable to 60,000+ for 1+ hour audio)

How It Works

1. User inputs text → GPT-4.1 generates a script
2. Script is processed line-by-line with MoonCast
3. Text is converted to semantic tokens
4. High-quality audio is generated with speaker consistency

🚀 Quick Start

Frontend Setup

1. Install Dependencies

   pnpm install

2. Set Up Environment

   • Sign up for AI provider accounts (OpenAI, Anthropic, etc.)
   • Obtain API keys
   • Copy .env.example to .env and fill in your API keys

3. Create Python Environment

   virtualenv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate
   pip install -r requirements.txt

4. Launch Development Server

   pnpm dev

MoonDIA TTS Setup

Prerequisites

• Conda installed on your system
• CUDA-compatible GPU (recommended)
• At least 8GB GPU memory

Step-by-Step Installation

1. Create and Activate Conda Environment

   conda env create -f environment.yml
   conda activate mooncast

2. Install MoonCast Dependencies

   cd MoonCast/
   pip install -r requirements.txt
   pip install flash-attn --no-build-isolation
   pip install huggingface_hub
   pip install gradio==5.22.0

   ⏱️ Note: flash-attn installation can take up to 5 hours

3. Download Pre-trained Models

   python download_pretrain.py

4. Set Up MoonDIA

   cd ../MoonDIA/
   
   # Copy resources from MoonCast
   cp -r ../MoonCast/resources/ CustomBuild/
   
   # Install additional requirements
   cd trained_mapper/
   pip install -r requirements_mooncast_2wice.txt
   pip install -r requirements_seq2seq.txt

5. Configure Environment

   # Create and configure .env file in trained_mapper directory
   cp .env.example .env
   # Edit .env with your API keys and configuration

🎵 Using the TTS System

The main TTS code is located in MoonDIA/trained_mapper/ with three main scripts:

Available Scripts

Script	Description	Usage
MoonCast_seed.py	Generates audio with 2 consistent speakers	python MoonCast_seed.py --input-file <file> --duration 5
MoonCast_no_prompt.py	Generates audio with random speakers throughout	python MoonCast_no_prompt.py --input-file <file> --duration 5
MoonCast_seed_explainer.py	Generates audio with 2 speakers + explanations	python MoonCast_seed_explainer.py --input-file <file> --duration 5

Example Usage

cd MoonDIA/trained_mapper/
python MoonCast_seed.py --input-file script.txt --duration 10

🔧 Technical Details

Architecture

• Frontend: Modern web UI with AI provider integration
• TTS Engine: MoonCast-based semantic token generation
• AI Integration: GPT-4.1 for intelligent script processing
• Optimization: 10-turn sliding window for speaker consistency

Performance Notes

• Current token limit: 15,000 (expandable to 60,000+)
• GPU memory requirement: 8GB+ recommended
• Processing time varies based on input length and GPU capability

📁 Project Structure

edtech/
├── README.md
├── environment.yml
├── requirements.txt
├── .env.example
├── MoonCast/
│   ├── resources/
│   ├── requirements.txt
│   └── download_pretrain.py
└── MoonDIA/
    ├── CustomBuild/
    │   └── resources/  # Copied from MoonCast
    └── trained_mapper/
        ├── MoonCast_seed.py
        ├── MoonCast_no_prompt.py
        ├── MoonCast_seed_explainer.py
        ├── requirements_mooncast_2wice.txt
        ├── requirements_seq2seq.txt
        └── .env

🤝 Contributing

This project demonstrates advanced TTS capabilities with local processing. The frontend serves as a design reference for future integration.

📝 License

This project is part of an educational technology initiative.

---