TurtleTalk - Indigenous Language Learning Platform

# TurtleTalk - Indigenous Language Learning Platform

An AI-powered language learning platform designed to preserve and revitalize Indigenous languages through interactive technology, focusing on Plains Cree with plans to expand to other Indigenous languages.

## 🎯 Project Overview

TurtleTalk combines modern web technologies with AI to create an accessible, culturally-sensitive platform for Indigenous language education. The platform features pronunciation evaluation, interactive lessons, community engagement, and cultural storytelling.

## ✨ Key Features

### Learning Platform
- **AI Pronunciation Evaluation**: Real-time feedback using Google Gemini API with 0-100 accuracy scores
- **Interactive Lessons**: Duolingo-style learning flow with multiple exercise types
- **Neural Text-to-Speech**: VITS-generated natural-sounding Indigenous language audio
- **Smart Progress Tracking**: Adaptive learning based on user performance
- **Audio Recording**: Browser-based voice capture for pronunciation practice

### Community Features
- **Cultural Stories**: Interactive storytelling in Cree with translations
- **Language Exchange**: Connect with native speakers and learners
- **Community Forums**: Discussion spaces for language and culture
- **Translation Collaboration**: Community-driven translation projects
- **Cultural Events**: Virtual and in-person Indigenous cultural events

### Accessibility
- **Screen Reader Support**: Full ARIA compliance for visually impaired users
- **Keyboard Navigation**: Complete keyboard-only navigation support
- **Adjustable Text Size**: Customizable font sizes for better readability
- **High Contrast Mode**: Enhanced visibility options
- **Audio Descriptions**: Comprehensive audio feedback for all interactions

## 🛠 Technical Implementation

### Frontend Architecture
- **React 18** with TypeScript for type-safe component development
- **Material-UI v5** for modern, accessible UI components
- **Web Audio API** for audio recording and playback
- **MediaRecorder API** for browser-based audio capture
- **React Router** for client-side navigation
- **Context API** for centralized state management

### Backend System
- **FastAPI** for high-performance Python web framework
- **SQLAlchemy ORM** for database operations and migrations
- **SQLite/PostgreSQL** for relational data storage
- **Google Gemini API** for multimodal AI pronunciation analysis
- **JWT Authentication** for secure user sessions
- **Pydantic** for robust data validation

### Machine Learning Services
- **VITS (Variational Inference with adversarial learning for Text-to-Speech)** for Indigenous language speech synthesis
- **Google Gemini 1.5 Flash** for context-aware pronunciation evaluation
- **Audio Processing** for WAV file generation and manipulation
- **Speech-to-Text Prototype** (completed) for voice recognition integration

## 💡 Technical Challenges Solved

1. **Indigenous Language TTS**: Implemented VITS neural network to generate natural-sounding speech for Plains Cree, a language with limited digital resources
2. **Real-time Pronunciation Evaluation**: Integrated Google Gemini API to provide instant, accurate feedback on user pronunciation
3. **Accessible Audio Recording**: Built browser-based audio capture system compatible with screen readers and keyboard navigation
4. **Cultural Sensitivity**: Designed UI/UX with input from Indigenous community members to ensure respectful representation
5. **Progressive Learning System**: Developed adaptive course modules that adjust difficulty based on user performance

## 🎬 User Experience Flow

1. **User Registration**: Create account with profile customization
2. **Course Selection**: Choose from available Indigenous language courses
3. **Interactive Lessons**: Complete exercises with listening, speaking, and translation
4. **Pronunciation Practice**: Record voice and receive AI-powered feedback with specific improvement tips
5. **Progress Tracking**: View achievements, scores, and learning analytics
6. **Community Engagement**: Participate in forums, cultural stories, and language exchange

## 📊 Project Statistics

- **20+ Git Commits** tracking iterative development
- **TypeScript**: 55.8% of codebase
- **Python**: 31.9% of codebase
- **JavaScript**: 9.6% of codebase
- **Full-stack architecture** with separate frontend, backend, and ML services

## 🚀 Future Enhancements

- **Whisper AI Integration** for conversational AI tutor with natural dialogue
- **Speech-to-Text Model** for real-time voice recognition (prototype completed)
- **Multi-language Support** (Ojibwe, Mohawk, Inuktitut, etc.)
- **Mobile Native Apps** for iOS and Android
- **Live Video Sessions** with native speakers
- **Gamification** with leaderboards and challenges
- **AI-generated Personalized Learning Paths**

## 🌍 Impact

TurtleTalk addresses the critical need for Indigenous language preservation through accessible digital technology. By combining AI-powered learning tools with cultural sensitivity and community engagement, the platform empowers both learners and native speakers to keep these vital languages alive for future generations.

## 🔗 Project Links

- **GitHub Repository**: [TurtleTalk](https://github.com/Mayalevich/TurtleTalk)
- **Documentation**: Architecture specs, API contracts, and setup guides available in repository
- **Tech Stack**: React, TypeScript, FastAPI, Python, Google Gemini AI, VITS TTS, Material-UI

---

*Built with 🐢 for HackHive 2026 - Made for Indigenous language preservation*