I-pole-star/pronouncepath-backend

PronouncePath

PronouncePath is a full-stack gamified web application designed to help users improve their pronunciation, speech fluency, and phonetics. Built with a modern tech stack, it evaluates real-time microphone recordings against target sentences, tracks user progression through tiers, and provides detailed word-by-word feedback.

Key Features

1. 1.Speech Evaluation: Uses machine learning models (Wav2Vec2) to analyze audio and score pronunciation accuracy.
2. 2.Gamified Progression: Earn XP as you practice, unlock new tiers (Beginner , Intermediate , Advanced), and track your growth.
3. 3.Phonetic Breakdown: Visual word-by-word comparison using G2P (Grapheme-to-Phoneme) mapping to show which sounds were correct or missed.
4. 4.Smart Audio Guard: Automatically detects and blocks blank recordings or background silence to prevent false scoring.
5. 5.Attempt History: Logs past practice sessions so users can review their progress over time.

Tech Stack

• —Frontend: React, Vite, JavaScript, CSS
• —Backend: FastAPI, Python, PyTorch, SQLAlchemy
• —Database: SQLite
• —AI/NLP: Hugging Face Transformers (Wav2Vec2), g2p_en, PyAV (Audio Processing)