clearmind pro

💡 Inspiration

As students, we’ve all experienced "textbook blindness"—reading dense paragraphs 5 times without retaining a single concept. Standard chatbots often dump another generic wall of text without diagnosing why a student is confused. Furthermore, in multilingual households across the world (especially in India where millions think and learn in Hinglish), language barriers make technical subjects even more intimidating.

We built ClearMind Pro to act like the ultimate empathetic human mentor: an AI tutor named Luna who speaks your language, listens via real-time Socratic voice, analyzes textbook photos using multimodal vision OCR, generates instant 60-second exam cheat sheets, and gamifies mastery with a rapid-fire Blitz battle arena.

🚀 What ClearMind Pro Does

ClearMind Pro is an interactive, obsidian-dark split-screen learning ecosystem with two synchronized halves:

1. Conversational AI Tutor (Luna):

   • Teaches any subject (Physics, Calculus, Biology, Chemistry, Coding, Literature, etc.) calibrated to the student's exact level (Class 1–12, College, or Competitive Exams).
   • Explains concepts through vivid physical metaphors and Interactive Analogy Cards.
   • Supports 8+ languages including natural conversational Hinglish (Hindi + English).
   • Real Multimodal Vision OCR: Students can snap or upload a photo of dense textbook formulas or diagrams (📷), and Luna reads and explains them step-by-step.

2. Synchronized Live Knowledge Canvas:

   • Interactive Learning Roadmap: Automatically charts step-by-step curriculum milestones for any commanded topic, allowing students to click any milestone to start a focused lesson.
   • 60-Second Exam Cheat Sheet: Real-time synthesis of #1 examiner traps, golden formulas, S.P.A.R.K. memory mnemonics, and guaranteed 5-mark exam questions, with 1-click isolated printing.
   • 60-Second Rapid-Fire Blitz Arena: High-intensity 1-tap MCQ battle against a ticking clock with combo multipliers, score tallies, and strict answer-locking when time expires.
   • Live Voice Orbit: Hands-free two-way conversational voice call using Microsoft Edge Neural Voice and browser speech recognition.
   • Student Journey & Gamification: Dynamic XP progression, daily study streaks, and level milestone badges.

🔧 How We Built It

• Backend Architecture: Built with FastAPI (asynchronous Python) for ultra-low latency response streaming and neural voice audio delivery.
• Multimodal AI Reasoning: Powered by Google Gemini 3.6 Flash (with zero-downtime GLM-4 failover) executing multimodal vision OCR, structured schema generation, and conceptual analogy synthesis.
• Neural Voice Synthesis: Integrated Microsoft Edge Neural TTS (edge-tts) for human-like expressive voice narration, with fallback to browser SpeechSynthesis.
• Scientific Typography: Embedded KaTeX for mathematical and scientific formula typesetting ($E=mc^2$, calculus derivatives, matrices).
• Obsidian Dark UI: Handcrafted responsive layout with Tailwind CSS, dynamic drag-and-drop splitter panes, glassmorphic panels, and Web Audio API synthesized sound effects.

🥊 Challenges We Ran Into

• Natural Conversational Hinglish: Traditional translation engines produce robotic, overly formal Hindi. We engineered prompt personas that blend natural conversational Hindi and English while preserving English technical terms.
• Multimodal Token Synchronization: Parsing uploaded textbook photos alongside complex math prompts required structured base64 decoding and strict Pydantic model validation.
• KaTeX Escaping in Dynamic Markdown: Standard markdown parsers escaped KaTeX span classes into raw HTML entities. We implemented a tokenized placeholder pipeline to ensure all equations render with textbook-level clarity.

🏅 Accomplishments That We're Proud Of

• ⚡ 1-Click Startup: Created a zero-friction batch script (run.bat) that launches the server and opens the browser in under 2 seconds.
• 👁️ Full Multimodal Loop: Real textbook image OCR with step-by-step mathematical reasoning.
• 🗺️ Instant Roadmap Synchronization: Commands like "create a roadmap for relations and functions" dynamically populate the visual canvas.
• ⏱️ Cheating-Proof Quiz Arena: Complete 60s countdown with answer locking and combo multiplier sound FX.

📚 What We Learned

• Multimodal learning (combining visual analogies, voice playback, and active quiz testing) creates significantly higher concept retention than passive chatbot text.
• Clean glassmorphism UI with gamification (XP & streaks) transforms intimidating technical topics into an engaging, enjoyable routine.

🔮 What's Next for ClearMind Pro

• 📱 Native Mobile App with real-time AR textbook camera solving.
• 👥 Collaborative Multiplayer Blitz Battles where classmates can challenge each other live.
• 📊 Educator Misconception Heatmaps allowing teachers to pinpoint where their class loses marks.

Built With

• edge-tts
• fastapi
• google-gemini
• html5
• javascript
• katex
• multimodal-ai
• python
• tailwind-css
• web-speech-api