AI Debate Generator: Wikipedia → Hinglish Debates with Natural TTS | RadioBot

*AI-Powered Debate Generator: Turn Wikipedia Articles Into Natural Hinglish Debates!*

In this video, I demonstrate RadioBot - an AI system that automatically generates realistic, conversational debates in Hinglish (Hindi-English mix) from any Wikipedia article. Watch as the system downloads articles, extracts debate topics, and creates natural-sounding dialogue between two speakers with proper pauses, emphasis, and even interruptions!

---

🎯 What You'll Learn

✅ How to build an AI debate generation pipeline using RAG (Retrieval-Augmented Generation)
✅ Integrating multiple LLM providers (Groq for speed, Google Gemini for quality)
✅ Generating natural Hinglish conversations with proper grammar
✅ Implementing SSML formatting for realistic text-to-speech
✅ Creating debate structures with openings, rebuttals, and closings
✅ Converting text to speech with multiple TTS providers

---

🔧 Technology Stack

**RAG Pipeline**: ChromaDB + Sentence Transformers for context retrieval
**LLM Providers**: Groq (topic generation) + Google Gemini (debate generation)
**TTS Providers**: ElevenLabs, OpenAI, Google TTS, Bark
**Languages**: Python, LangChain, ChromaDB
**Features**: SSML support, natural interruptions, multi-speaker audio

---

🚀 Key Features Demonstrated

1. **Automatic Topic Extraction**: AI identifies debatable topics from Wikipedia content
2. **RAG-Enhanced Context**: Retrieves relevant information for informed debates
3. **Natural Hinglish Dialogue**: Generates conversational Hinglish with proper grammar
4. **SSML Support**: Includes pauses, emphasis, and prosody for natural TTS
5. **Multi-Provider TTS**: Supports OpenAI, ElevenLabs, Google, and Bark
6. **Interruptions**: Models natural speech interruptions and responses
7. **Complete Pipeline**: From Wikipedia article to final MP3 audio file

---

💻 Code & Resources

**GitHub Repository**: https://github.com/Appurava/RadioBot

*Key Files:*
`main.py` - Main entry point
`DebateGenerator.py` - LLM debate generation
`EncrichmentPipeline.py` - Prompt enrichment
`TextToSpeech.py` - TTS conversion


**Documentation**: Full design documentation and test suite included

---

🎓 Technical Deep Dive

*RAG Pipeline:*
Downloads Wikipedia articles
Chunks documents with overlap
Generates embeddings using Sentence Transformers
Stores in ChromaDB vector database
Retrieves relevant context for debate generation

*LLM Integration:*
Groq API for fast topic extraction
Google Gemini for high-quality debate generation
Retry logic with exponential backoff
Google API retry delay extraction

*SSML Formatting:*
`break time="X.Xs"/` for natural pauses
`emphasis level="strong"` for important words
`prosody rate="slow/fast"` for speech variation
Full W3C SSML standard compliance

---

🔑 API Keys Required

Groq API Key (free tier available)
Google API Key (free tier available)
ElevenLabs API Key (for TTS - optional)
OpenAI API Key (for TTS - optional)

---

📚 Related Topics

#AIDebate #Hinglish #TextToSpeech #RAG #LLM #PythonProject #AIProject #MachineLearning #NaturalLanguageProcessing #ChromaDB #Groq #GoogleGemini #ElevenLabs #SSML #Wikipedia #DebateGeneration #ConversationalAI #IndianEnglish #CodeTutorial #AITutorial

---

💬 Connect
LinkedIn: https://www.linkedin.com/in/appurava/

---

⚠️ Important Notes

This project requires API keys for LLM and TTS providers
Some providers have usage limits on free tiers
Audio quality depends on TTS provider selected
SSML support varies by TTS provider

---

🙏 Credits

Built using:
LangChain for document processing
ChromaDB for vector storage
Groq & Google for LLM services
ElevenLabs, OpenAI, Google for TTS

---

*Like this video?* Hit the 👍 button and subscribe for more AI project tutorials!

*Questions?* Drop them in the comments below!#aidebates , #Hinglish, #TextToSpeech, #RAG, #LLM, #PythonProject, #AIProject