Lyric-based Music Recommendation System 🎵

Project Title: J.O.H.A.N.N (joˈhan)

Just One Hip Algorithm for Navigating Narratives

---

Overview

The Lyric-based Music Recommendation System, J.O.H.A.N.N, aims to provide personalized music recommendations by analyzing the sentiment of song lyrics. Utilizing the Spotify API to gather a user's saved songs, web-crawling Genius.com to extract lyrics, and employing GPT sentiment analysis, the system suggests songs with matching emotional tones.

---

Table of Contents

• Background
• Design Process
• Features
• Usage
• Technologies Used

---

Background

Have you ever wondered whether your friends pay more attention to the lyrics or the instrumental part of a song? Many focus on the beat, leaving lyric enthusiasts like me in the dark when it comes to recommendation algorithms. This project addresses that gap by creating a sentiment-driven music recommendation system that combines data from the Spotify API, Genius.com, and ChatGPT.

---

Design Process 🛠️

Current stage of development:

• 12 & 13. Tests & Documentation

---

Features

• Spotify Integration: Fetches a user's saved songs from Spotify to create a personalized music library. 🎧
• Lyrics Extraction: Web-crawls Genius.com to extract lyrics for each song in the library.
• Sentiment Analysis: Analyzes the sentiment of song lyrics to determine emotional tones. ❤️
• Playlist Generation: Generates a playlist title and description based on the cluster of songs.
• Playlist Picture: Generates a playlist picture based on the title and description.
• Recommendation Engine: Provides personalized song recommendations based on similar sentiment analysis on a single song. 🔍

---

Usage

1. Install dependencies:

   pip3 install -r requirements.txt

2. Set up Spotify API credentials:

   • Visit the Spotify Developers page. Create an account, a project (Spotify calls it an 'app'), and find the client ID & client secret in 'settings'.

3. Set up Postgres:

   • Use the provided schema file.

4. Set up .env file:

   • Follow the instructions in the .env file.

5. Run the webcrawler:

   python3 webcrawl_lyrics.py

6. Authenticate Spotify and initiate the recommendation process:

   • Follow the on-screen instructions.

7. Run the summary making program:

   python3 summary.py --generate_batch

   • Check the status at any time:

   python3 summary.py --check_status

   • Note: This checks status every 10 min. Use 'ctrl+c' to stop manually.

8. Run the embeddings:

   python3 embeddings.py

9. Run the mapping with HDB or K-Means:

   python3 mapping.py --dense <CLUSTER_MIN_SIZE_INT> <CLUSTER_MIN_SAMPLES_INT>

   • Example:

   python3 mapping.py --dense 10 4

   • Note: Run multiple times to dial in the right clusters.

10. Create the playlists:

python3 make_cluster_playlists.py

---

Tech Stack

Database: Postgres 🗄️

Program: Python 🐍

---

Technologies Used

• Spotify API 🎵
• Selenium (for dynamic web crawling) 🌐
• Beautiful Soup (for static web crawling) 🍲
• Clustering algorithms 📊
• ChatGPT API 🤖