calm-speech

bySamarth Choudhar

You are a senior full-stack AI engineer. Build a complete production-ready project called "AI Voice English Trainer". 📌 Project Goal: A web application where users speak English and get real-time feedback on: - Grammar - Pronunciation - Fluency - Suggestions for improvement --- 🧩 FEATURES REQUIRED: 1. Speech Input: - Capture user voice using browser (Web Speech API or MediaRecorder) - Convert speech to text 2. Grammar Correction: - Use an LLM API (OpenAI or Gemini) - Return: - Corrected sentence - Explanation of mistakes - Better alternative sentences 3. Pronunciation Feedback: - Compare spoken text vs ideal text - Provide score (0–100) - Highlight incorrect words 4. Feedback System: Return structured output: { "original_text": "", "corrected_text": "", "grammar_feedback": "", "pronunciation_score": "", "suggestions": [] } 5. Daily Practice Tracking: - Store user history (local storage or backend DB) - Show previous attempts --- 🖥️ FRONTEND: - HTML, CSS (Glassmorphism UI), JavaScript - Pages: - Home - Practice Page (Record + Results) - History Page - Features: - Record button 🎤 - Live transcript - Feedback display cards - Progress UI --- ⚙️ BACKEND: - Python FastAPI - API endpoints: - /speech-to-text - /analyze-text - /history - Use: - OpenAI/Gemini API for grammar correction - Basic scoring logic for pronunciation --- 🧠 AI INTEGRATION: - Prompt LLM like: "Correct the grammar and explain mistakes in simple English. Suggest better alternatives." --- 📁 PROJECT STRUCTURE: frontend/ index.html style.css script.js backend/ main.py routes/ services/ requirements.txt --- ✨ EXTRA FEATURES (IMPORTANT): - Loading animation - Error handling - Clean UI - Mobile responsive --- 🚀 OUTPUT REQUIRED: 1. Full working code (frontend + backend) 2. Step-by-step setup instructions 3. API key integration guide 4. Sample test cases 5. GitHub README content --- 🎯 IMPORTANT: - Keep code beginner-friendly but clean - Use comments to explain logic - Make it deployable Now generate the full project.