Introduction

Enchanted Shooter is a bubble shooter–style lyric guessing game that uses an N-gram language model trained on Taylor Swift lyrics. Built with Svelte (frontend) and FastAPI (backend), the game combines interactive design and NLP to create an engaging educational + entertainment experience.

Gameplay Flow

1. Song Selection – Player chooses a Taylor Swift song.

2. Difficulty Level – Choose between easy, medium, or hard (controls the lyric gaps and N-gram challenge).

3. Game Start – 10 lyric questions per round.

4. Hints – Up to 3 hints can be used (based on N-gram predictions).

5. Scoring – Starts at 100 points.

• Correct Answer: Full points retained.
• Wrong Answer: -10 points each.
• Maximum score: 100, minimum: 0.

Sample Screenshots

< >

N-Gram Integration

• Corpus: Taylor Swift lyrics.
• Function: Generates predictions for missing words based on N-gram frequencies.
• In-Game Role: Provides hints and influences difficulty.
• Example: If the lyric is “I knew you were ___”, the model predicts likely completions (e.g., “trouble”).

System Architecture

• Frontend (Svelte) – Bubble shooter interface, lyric display, hint requests, score tracking.
• Backend (FastAPI + Python) – Hosts N-gram model, processes requests, returns predictions, and validates answers.
• Flow: Player input → API request → N-gram processing → Prediction/validation → Game update.

Features

• Bubble shooter + lyric guessing combo.
• Difficulty levels that scale with N-gram complexity.
• Hint system powered by NLP predictions.
• Fair scoring with penalties for mistakes.