Simplified And Automated Research Amplification and Learning
Transform research papers into educational videos, podcasts, mind maps, and visual stories using AI.
Quick Links: Live Demo | Chrome Extension | WhatsApp Bot | Contact
- Overview | Features | Installation | Configuration | Running | Chrome Extension | WhatsApp Bot | Troubleshooting | API Docs | Contributing | License
Research Paper β AI Processing β πΉ Video | ποΈ Podcast | πΊοΈ Mindmap | π Story
π Chrome Extension: Process papers from any website!
Key Capabilities:
- π₯ Educational videos with AI narration
- ποΈ Natural podcast conversations
- πΊοΈ Visual mind maps
- π Cinematic storytelling videos
- π Chrome extension for instant processing
- π¬ WhatsApp bot for 24/7 research assistance
- π Multi-language (English/Hindi/Gujarati)
| Feature | Description |
|---|---|
| Video Generation | AI scripts, professional slides, multi-language narration |
| Podcast Creation | Two-voice dialogues, natural conversations |
| Mind Mapping | Hierarchical concept visualization, SVG export |
| Visual Stories | Scene-by-scene narratives with text overlays |
| Chrome Extension | One-click processing from arXiv, Google Scholar |
| WhatsApp Bot | 24/7 research assistance via chat |
| Batch Processing | Handle multiple papers efficiently |
Students: Exam prep, quick paper understanding, visual learning
Educators: Lecture content, teaching materials, multi-format resources
Researchers: Conference presentations, accessible findings, outreach
Institutions: Content libraries, online courses, research accessibility
Chrome Extension: Process papers instantly from any research website
WhatsApp Bot: Chat-based research assistance, anywhere, anytime
π Tutorial: Download PDF for complete setup and usage guide
Backend: Python 3.9+, LaTeX (pdflatex/MiKTeX), Poppler, FFmpeg, 4GB+ RAM
Frontend: Node.js 16+, npm 8+, Modern browser
API Keys: Google Gemini (required, free 200 req/day), Sarvam AI (optional, Hindi TTS), Hugging Face (optional, free AI images)
GGW_Megathon_Saral/
βββ chrome-extension/ # Browser extension (manifest.json, popup, content scripts)
βββ backend/ # FastAPI server (routes, services, models)
β βββ app/ # Main application (routes/, services/, models/)
β βββ temp/ # Generated files (papers, videos, podcasts, mindmaps)
βββ frontend/ # React app (pages, components, contexts)
Related Repository:
βββ Research-Paper-Chatbot/ # WhatsApp bot companion (Twilio + Gemini)
βββ https://github.com/N1KH1LT0X1N/Research-Paper-Chatbot
Windows: Python, Node.js, MiKTeX, Poppler, FFmpeg
macOS: brew install python@3.9 node poppler ffmpeg && brew install --cask mactex
Linux: sudo apt install python3.9 nodejs npm poppler-utils ffmpeg texlive-full
# Clone repository
git clone https://github.com/yourusername/GGW_Megathon_Saral.git
cd GGW_Megathon_Saral
# Backend
cd backend
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install -r requirements.txt
# Frontend
cd ../frontend
npm installCreate .env in backend/ directory:
GEMINI_API_KEY_1=AIzaSy... # Get from https://aistudio.google.com/apikey
GEMINI_API_KEY_2=AIzaSy... # Optional: for quota rotation
SARVAM_API_KEY=your_key # Get from https://www.sarvam.ai/
HUGGINGFACE_API_KEY=hf_... # Optional: https://huggingface.co/settings/tokensAPI Key Rotation: Add multiple Gemini keys (GEMINI_API_KEY_1, _2, etc.) for automatic rotation when quota limits hit.
Web UI Setup: Configure keys through the API Setup page after launching the app.
Backend (Terminal 1):
cd backend && source .venv/bin/activate # Windows: .venv\Scripts\activate
uvicorn app.main:app --reload --host 0.0.0.0 --port 8000β Backend: http://localhost:8000 | API Docs: http://localhost:8000/docs
Frontend (Terminal 2):
cd frontend && npm startβ Frontend: http://localhost:3000
Video Generation: Upload paper β Generate script β Edit content β Assign images β Generate video (English/Hindi, adjustable complexity)
Podcast Creation: Upload paper β Generate dialogue β Customize voices β Create MP3 (natural two-voice conversation)
Mind Mapping: Upload paper β AI extracts concepts β Download SVG (hierarchical structure, relationships)
Visual Story: Upload paper β Generate scenes β Create video with text overlays β Watch/download
Chrome Store: Coming soon
Manual: chrome://extensions/ β Enable Developer mode β Load unpacked β Select chrome-extension folder
- Right-click any paper on arXiv/Google Scholar β "Generate Video/Podcast with SARAL"
- Extension popup β Detects paper β One-click generation
- Highlight URL β Right-click β "Process with SARAL"
Auto-detects papers on arXiv, Google Scholar, PubMed, ResearchGate, IEEE, Springer, ScienceDirect | One-click generation | Batch processing | Real-time notifications | Keyboard shortcut: Ctrl+Shift+S
Configure backend URL (http://localhost:8000), complexity level, language, theme via extension popup
Your 24/7 AI research assistant for semantic search, Q&A, and summarization of research papers via WhatsApp.
Join the Bot: WhatsApp Link
Repository: Research-Paper-Chatbot
Live Demo: https://research-paper-chatbot-2.onrender.com
| Feature | Description |
|---|---|
| Semantic Search | Natural language search across research papers |
| Q&A System | Ask questions about specific papers and get scholarly answers |
| Paper Summaries | Get concise summaries of research papers |
| Topic Explanations | Simple explanations suitable for different learning levels |
| Activities Generator | Generate educational activities based on research topics |
Send messages to the WhatsApp bot:
# Search for papers
"transformer attention"
"retrieval augmented generation"
# Select a paper
"select 1"
# Start Q&A session
"ready for Q&A"
"tell me more about attention is all you need"
# Get explanations
"Explain transformers"
# Generate activities
"Activities machine learning"
- Backend: Flask + Python 3.9+
- AI Engine: Google Gemini API
- Messaging: Twilio WhatsApp API
- Search: Semantic search with arXiv integration
- PDF Processing: PyMuPDF for paper extraction
If you want to deploy your own instance:
# Clone the bot repository
git clone https://github.com/N1KH1LT0X1N/Research-Paper-Chatbot.git
cd Research-Paper-Chatbot
# Install dependencies
pip install -r requirements.txt
# Configure environment
# Create .env file with:
# TWILIO_ACCOUNT_SID=your_sid
# TWILIO_AUTH_TOKEN=your_token
# GEMINI_API_KEY=your_key
# Run the bot
python research_bot.py
# Expose with ngrok for webhook
ngrok http 5000Webhook Setup: Configure Twilio WhatsApp sandbox webhook to point to https://your-ngrok-url.ngrok.io/whatsapp
The WhatsApp bot complements the SARAL AI web platform by providing:
- Mobile-First Access: Research assistance on-the-go without browser
- Conversational Interface: Natural language interactions
- Quick Queries: Instant answers without full video generation
- Accessibility: Chat-based interface for users with limited connectivity
Use the web platform for comprehensive content generation (videos, podcasts, mindmaps) and the WhatsApp bot for quick research queries and paper discovery.
Common Issues:
- ImportError β
pip install -r requirements.txtin venv - PDF/LaTeX errors β Install poppler, MiKTeX/texlive, add to PATH
- FFmpeg not found β Install and add to PATH
- API key invalid β Check
.envformat (no quotes:KEY=value) - Gemini quota β Add multiple keys:
GEMINI_API_KEY_1,_2, etc. - Port in use β Kill process or change port
- npm install fails β Delete
node_modules, reinstall - No audio in video β Verify Sarvam API key
- Extension issues β Reload from
chrome://extensions/ - WhatsApp bot not responding β Check Twilio webhook configuration and API keys
Get Help: GitHub Issues | Email | WhatsApp Bot
Tech Stack:
Frontend: React 18.x, Tailwind CSS, Framer Motion, React Router, Axios
Backend: FastAPI, Google Gemini API, Sarvam AI, MoviePy, FFmpeg, PyMuPDF, PIL
Testing: npm test (frontend), pytest (backend)
Customization: Edit tailwind.config.js for themes, TTS service for voices, slide_generator.py for video templates
Base URL: http://localhost:8000/api
Key Endpoints: /papers/upload, /papers/arxiv, /scripts/generate, /slides/generate, /media/generate-audio, /media/generate-video, /podcast/generate, /mindmap/generate, /visual-storytelling/generate-storytelling-script, /visual-storytelling/generate-storytelling-video
Interactive Docs: http://localhost:8000/docs - Complete reference, testing, schemas
Report Bugs: Open issue with description, steps to reproduce, error logs
Suggest Features: Open issue with use case and benefits
Submit PRs: Fork β Create branch β Commit β Push β Open PR
Code Style: Python (PEP 8), JavaScript (ESLint), Conventional commits
This project consists of two complementary repositories:
- SARAL AI Platform (This Repo) - Full-featured web application
- WhatsApp Bot - Mobile-first chat interface
Both repositories are maintained by the SARAL AI team and work together to democratize research access.
MIT License Β© 2025 SARAL AI Team - See LICENSE for details
AI & APIs: Google Gemini, Sarvam AI, Hugging Face
Frameworks: FastAPI, React, Tailwind CSS, MoviePy, FFmpeg
Tools: arXiv, LaTeX, Poppler
Email: democratise.research@gmail.com
WhatsApp Bot: Join Bot
GitHub Issues: Report Bugs
Bot Repository: Research-Paper-Chatbot
β Star this repository if you found it helpful!
Made with β€οΈ by the GitGoneWild Team | Making Research Accessible to Everyone